United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT
_________
Argued October 24, 2005 Decided January 27, 2006
No. 04-5414
NOVARTIS PHARMACEUTICALS CORPORATION,
APPELLANT
v.
MICHAEL O. LEAVITT,
SECRETARY OF HEALTH AND HUMAN SERVICES, AND
LESTER M. CRAWFORD, JR.,
ACTING COMMISSIONER OF FOOD AND DRUGS,
APPELLEES
________
Appeal from the United States District Court
for the District of Columbia
(No. 99cv00323)
_________
John M. Engel III argued the cause and filed the briefs for
appellant.
Drake Cutini, Attorney, U.S. Department of Justice,
argued the cause for appellees. With him on the brief were
Peter D. Keisler, Assistant Attorney General, Eugene M.
Thirolf, Jr., Director, Alex M. Azar II, General Counsel, U.S.
Department of Health and Human Services, Eric M.
2
Blumberg, Deputy Chief Counsel, and Karen E. Schifter,
Associate Chief Counsel.
Before: TATEL and GRIFFITH, Circuit Judges, and
WILLIAMS, Senior Circuit Judge.
Opinion for the Court filed by Senior Circuit Judge
WILLIAMS.
WILLIAMS, Senior Circuit Judge: Securing FDA
approval for a generic drug is generally a much simpler and
faster process than securing approval for a “pioneer” drug.
Instead of directly demonstrating the drug’s safety and
efficacy, see 21 U.S.C. § 355(a), manufacturers of the generic
file an abbreviated new drug application (“ANDA”) that need
show only that the generic is the same as the pioneer drug
along certain dimensions. See § 355(j)(2)(A)(i)-(viii). This
case concerns FDA’s decision to change the dosage forms,
labeling, and established names associated with appellant
Novartis’s pioneer drug in ways that would ease the path for
competing generic drugs. Novartis raises a number of
procedural and substantive challenges to FDA’s changes. The
district court rejected all and we affirm.
* * *
The fastest route to FDA approval requires an ANDA to
demonstrate not only that a generic drug has the same active
ingredient(s) as the pioneer drug and is bioequivalent to it,
i.e., roughly speaking, is absorbed at the same rate and to the
same extent when administered under similar conditions, but
also that the generic and pioneer drugs have the same dosage
form and labeling. 21 U.S.C. § 355(j)(2)(A)(ii)-(v). The
latter two requirements are at issue in this case.
3
“Dosage forms” are categories that relate to such matters
as a drug’s appearance, physical form, and method of
administration. The FDA assigns each drug a dosage form;
currently there are 77 such categories. See FDA, APPROVED
DRUG PRODUCTS WITH THERAPEUTIC EQUIVALENCE
EVALUATIONS Appx. C (25th ed. 2005). Some of the dosage
forms are quite broad (for example, “elixir” or “tablet,
chewable”), while others are relatively narrow (for example,
“injectable, lipid complex”). Id. If a generic manufacturer
cannot show that its drug has the same dosage form as the
pioneer drug, it can still obtain FDA approval, but the process
is more arduous: the manufacturer may use an abbreviated
application only if it first files a “suitability petition” and the
FDA grants it permission to file an ANDA. 21 U.S.C.
§ 355(j)(2)(C); 21 C.F.R. § 314.93. Even if the drug is
ultimately approved, the difference in dosage form will
preclude the generic from being designated therapeutically
equivalent to the pioneer drug, and will thus disqualify the
generic from being considered automatically substitutable for
the pioneer drug under various state pharmacy laws. See
Warner-Lambert v. Shalala, 202 F.3d 326, 327-28 (D.C. Cir.
2000).
To avoid charges of misbranding, the labels of
prescription drugs must include certain pieces of information
displayed in a particular way. See 21 U.S.C. § 352. Thus the
requirement that a generic have the same labeling as the
pioneer drug incorporates several additional requirements, two
of which are relevant to this case.1 First, FDA requires the
1
The same-label requirement excepts differences required
because the generic was approved under a suitability petition or
“because the new drug and the listed drug are produced or
4
labels to contain the pioneer drug’s dosage form, see 21
C.F.R. § 201.57(a)(1)(ii), so the same-label requirement
reincorporates the requirement that the generic drug share the
pioneer drug’s dosage form. Second, the FDA requires
prescription drug labels to include the drug’s established
name, see 21 C.F.R. § 201.57(a)(1)(i), which is a
nonproprietary name assigned to the drug by the FDA. The
rule mandating inclusion of the established nonproprietary
name on the label means that a generic drug must have the
same nonproprietary name as the pioneer drug for the FDA to
approve the generic drug through the ANDA process. We
turn later to the statutory framework for assigning such
names.
Novartis markets cyclosporine drug products that are
widely prescribed to prevent organ rejection in kidney, liver,
and heart transplants. (These drugs were developed by
Novartis’s predecessor, Sandoz Corp. For simplicity, we refer
to both corporations as Novartis.) It markets two types of
cyclosporine drugs under the proprietary names Sandimmune
and Neoral. Both are available as capsules and as solutions,
and both form emulsions when they come into contact with
aqueous liquids. The difference is the size of the droplets in
which the active ingredient is dispersed. Sandimmune forms
a macroemulsion and disperses cyclosporine in larger droplets
than does Neoral, which forms a microemulsion. The smaller
size of the droplets in the microemulsion allows Neoral’s
active ingredient—cyclosporine—to be absorbed more
quickly and efficiently in the gastrointestinal tract. See FDA
distributed by different manufacturers.” 21 U.S.C.
§ 355(j)(2)(A)(v); 21 C.F.R. § 314.94(a)(8)(iv).
5
Docket No. 96-P-0459, Response to Petition Filed by Novartis
Pharmaceuticals Corp., at 4 (Nov. 2, 1998) (“Petition
Response”). This difference means that the Sandimmune and
Neoral products are not bioequivalent (i.e., they are
bioinequivalent) and that a physician’s decision to switch a
patient from one to the other may require a different
prescription.
To highlight the differences between Sandimmune and
Neoral, the FDA initially incorporated the term
“microemulsion” into both Neoral’s established name and its
dosage form. When the FDA approved the Neoral products in
1995, it assigned them the established names “cyclosporine
oral solution for microemulsion” and “cyclosporine capsules
for microemulsion.” Petition Response at 5; see also Joint
Appendix 966-67. Shortly thereafter, the FDA created two
new dosage forms and assigned them to the Neoral products:
“capsule, microemulsion” and “solution, microemulsion.”
At a symposium in December 1997 a leading transplant
physician made a presentation about the development of a
generic version of Neoral that formed a microdispersion of
solid particles instead of a microemulsion. Alarmed, Novartis
filed a citizen petition in March 1998 requesting that the FDA
not approve any generic with a dosage form that was not
identical to Neoral’s, or at least that it require an applicant
seeking approval of such a drug to first file a suitability
petition. (Novartis had previously filed another citizen
petition, which the FDA denied in the same response as it
denied the March 1998 petition; Novartis hasn’t appealed
denial of the first petition.) In principle the petition was
simply a request that the FDA obey the law, though it could
be taken as implicitly urging the FDA not to delete
“microemulsion” from the dosage forms assigned to the
6
Neoral products. In fact the FDA did just what Novartis
sought to avoid.
Responding to the petition in November 1998, the FDA
announced that it would eliminate the microemulsion dosage
forms altogether. It explained that it had determined that
Neoral’s ability to form a microemulsion was not a function
of dosage form; rather than being an aspect of the “physical
recognition, appearance, dosing and manner of
administration,” the microemulsion-forming feature of Neoral
appeared only after the patient swallowed the product (or, in
the case of the oral solution, mixed it with another beverage to
make it more palatable) and was therefore a characteristic of
the product’s “formulation.” Petition Response at 12-13 &
n.14. There is no requirement that a generic drug have the
same formulation as its pioneer. Moreover, the FDA said,
elimination of the microemulsion dosage form would serve its
policy goal of encouraging the availability of generic
products. Id. at 14. Finally, it said that in light of its dosage
form ruling it had reexamined the Neoral products’
established names and determined that they should no longer
refer to microemulsion. The new established names would be
simply “cyclosporine capsules” and “cyclosporine oral
solution.” These names would apply to Sandimmune, to
Neoral, and to any approved generic versions of Neoral that
formed either microemulsions or microdispersions. To
communicate the differences between Sandimmune on the one
hand and Neoral and generic equivalents that form either
microemulsions or microdispersions on the other, FDA said it
would now require the label for Neoral and its generic
equivalents to include the term “MODIFIED.” Petition
Response at 17. This term, explained the FDA, “is
appropriate to alert physicians, pharmacists, and patients that
Neoral’s formulation presents a different bioavailability
7
profile than the Sandimmune formulation,” and “is broad
enough to encompass different, bioequivalent formulations of
cyclosporine (e.g., a microemulsion or microdispersion) and,
prominently displayed in bold type, the term may become
readily associated with the more bioavailable formulations of
cyclosporine.” Id.
The FDA issued this decision around the time that it
approved an ANDA for SangCya, a generic cyclosporine
solution manufactured by SangStat Medical Corp. SangStat
had submitted data purporting to show that SangCya was
bioequivalent to Neoral. SangCya differed from Neoral in
precisely the way that Novartis had anticipated a generic
competitor might: it formed a microdispersion of solid
particles instead of a microemulsion. Had the Neoral
products’ dosage forms, labeling, and established names
continued to refer to microemulsion, SangStat would not have
been able to secure FDA approval for SangCya by directly
filing an ANDA.
A few months later, in February 1999, Novartis filed suit
in district court raising challenges to the FDA’s approval of
the SangCya ANDA as well as to the FDA’s modifications to
Neoral’s dosage forms, labeling, and established names. The
district court found that the first of these challenges was moot:
While this litigation was before the district court, new
evidence came to light revealing that SangCya was not
bioequivalent to Neoral when properly administered with
apple juice instead of with chocolate milk, and the FDA
withdrew its approval of SangCya. (FDA reports that as of
the time it submitted its reply brief, there were eight approved
and not withdrawn generic equivalents for Neoral, half of
which form microemulsions. Br. for Appellee 11.
Presumably the other half form microdispersions.) We affirm
8
the district court’s conclusion that the withdrawal mooted
Novartis’s challenge to that approval for the reasons given by
the district court. Mem. Op. (Mar. 5, 2001) 5-9.
On subsequent cross-motions for summary judgment, the
district court rejected the challenges to the FDA’s disposition
of Novartis’s citizen petition. Mem. Op. (Sept. 9, 2004). We
affirm the district court’s rejection of each of Novartis’s
substantive challenges to the FDA’s decisions regarding
dosage forms, labeling, and established names of its Neoral
products. Because we have nothing to add to the district
court’s reasoning on these issues, we do not address them
here. We do address below Novartis’s challenges to the
completeness of the administrative record and to the
procedures employed by the FDA to change Neoral’s dosage
forms, labeling, and established names. In each case, we
affirm the district court.
* * *
The first objection that Novartis raises on appeal is that
the district court erred by failing to require production of the
“whole record” as required by the Administrative Procedure
Act. We review the district court’s refusal to supplement the
administrative record for abuse of discretion. See James
Madison Ltd. v. Ludwig, 82 F.3d 1085, 1095 (D.C. Cir. 1996).
There are actually two administrative records at issue in
this case. The first is the public docket for Novartis’s citizen
petition, the second is the record of the FDA’s approval of the
SangCya ANDA. The district court initially referred the
question of the records’ proper scope to a magistrate judge,
who refused Novartis’s requests to supplement the citizen
petition record with records from a group of prior proceedings
9
that Novartis claimed were relevant, finding that FDA did not
consider these records in addressing the citizen petition.
Mem. Op. (Jan. 18, 2000) 5-8. The magistrate judge did
provide Novartis with the SangCya ANDA approval record
after withholding some trade-secret, confidential-commercial,
and otherwise privileged information. Id. at 8-11. The district
court affirmed the magistrate judge’s determinations. Mem.
Op. (May 4, 2000).
When the new evidence concerning SangCya’s
bioinequivalence became available, Novartis again sought to
supplement the record. The magistrate judge agreed with
Novartis in part, expanding the record to include both the new
bioinequivalence data and the hitherto withheld portions of
SangCya’s ANDA relating to its asserted bioequivalence.
Mem. Op. (Nov. 27, 2000). In a subsequent decision the
district court addressed both SangStat’s motion to dismiss
Novartis’s challenge to the SangCya approval (by then
withdrawn) as moot and SangStat’s separate motion to
reconsider the magistrate judge’s determination regarding the
scope of the record. The district court found that Novartis’s
attack on the SangCya approval was moot, Mem. Op. (Mar. 5,
2001) 5-9; as noted above, we affirm that decision for the
reasons stated. The district court then vacated the magistrate
judge’s expansion of the record, regarding its decision on
substantive mootness as controlling that issue. Id. at 9.
On appeal, Novartis renews the supplementation requests
denied by the district court. In making its case Novartis relies
heavily on American Bioscience, Inc. v. Thompson, 243 F.3d
579 (D.C. Cir. 2001), and Walter O. Boswell Memorial Hosp.
v. Heckler, 749 F.2d 788 (D.C. Cir. 1984). These cases do
little to support Novartis’s position. The agency failed to file
any administrative record at all in the former case, American
10
Bioscience, Inc., 243 F.3d at 582, while in the latter neither
party purported to have provided the court with the complete
administrative record, Walter O. Boswell Memorial Hospital,
749 F.2d at 792. We find no abuse of discretion in the district
court’s holding that the administrative record was complete
and sufficient for judicial review; we adopt its reasoning
except with respect to Novartis’s efforts to include the data
that originally purported to establish SangCya’s
bioequivalence to Neoral. Rather than resting on the
mootness of Novartis’s challenges to SangCya’s approval, we
rely on the district court’s later finding that Novartis has
pointed to nothing to support its claim that the FDA’s
approval of SangCya tainted its consideration of the logically
distinct issues raised by Novartis’s own citizen petition. See
Mem. Op. (Sept. 9, 2004) 26-27. Novartis makes one other
claim about the record, but because it is closely related to
Novartis’s procedural challenge to the FDA’s modifications of
the Neoral products’ established names, we address it in the
next section.
* * *
Novartis raises two procedural objections that apply to
each of the FDA’s decisions regarding Neoral’s dosage forms,
labeling, and established names, and a third procedural
objection that applies only to the FDA’s modification of the
established names for the two Neoral products. These
challenges rely, in varying proportions, on interpretations of
the Federal Food, Drug, and Cosmetic Act (“FDCA”) and
regulations interpreting that statute. We have held on a
number of occasions that FDA interpretations of the FDCA
receive deference, as do its interpretations of its own
regulations unless plainly erroneous or inconsistent with the
11
regulations. See, e.g., Purepac Pharmaceutical Co. v.
Thompson, 354 F.3d 877, 883 (D.C. Cir. 2004); Mova
Pharmaceutical Corp. v. Shalala, 140 F.3d 1060, 1071 n.13
(D.C. Cir. 1998).
First, Novartis objects to the FDA’s failure to publicly
docket any of SangStat’s requests that the FDA modify the
dosage forms, labeling, and established names that apply to
Novartis’s Neoral products, saying that the FDA violated its
own procedures by not treating those requests as citizen
petitions. See 21 C.F.R. § 10.30. The district court held that
the provision permitting an interested person to petition the
FDA “does not require FDA to convert every letter or
telephone call it receives in conjunction with an ANDA into a
citizen petition.” Mem. Op. (Sept. 9, 2004) 24. And, the
district court pointed out, another regulation forbids the FDA
from disclosing information about a pending ANDA unless its
existence has previously been publicly disclosed. Id. (citing
21 C.F.R. § 314.430(b)-(c)). We affirm.
Second, Novartis insists that the FDA erred by failing to
provide Novartis with notice or opportunity to comment on its
decisions. The district court disagreed, pointing out that
Novartis’s own citizen petition gave it an opportunity to show,
in detail, why it believed that the microemulsion dosage forms
were appropriate. Mem. Op. (Sept. 9, 2004) 25. We also note
that Novartis acknowledged the link between dosage form and
labeling in that same petition and that the public docket
indicates Novartis took advantage of its opportunity to
respond to comments made by others in response to its
petition. As Novartis received ample notice and opportunity
to be heard, it has already received every benefit that it could
from a favorable judgment on this issue. Better Government
12
Ass’n v. Department of State, 780 F.2d 86, 91 & n.21 (D.C.
Cir. 1986).
Third, Novartis makes a more specific argument that the
FDA’s actions in modifying the Neoral products’ established
names were not procedurally proper. Our analysis of this
argument is more complicated because Novartis and the FDA
disagree on how to characterize the status of the
nonproprietary names assigned to the Neoral products when
the FDA approved them in 1995. Novartis argues that the
FDA designated “official” names pursuant to 21 U.S.C.
§ 358(a). This is significant, Novartis argues, because it
means that the FDA may not modify those names without
undertaking notice-and-comment rulemaking. We conclude
that Novartis fails to establish either that the FDA was
required to proceed under § 358, or that as a discretionary
matter it did proceed under that section, in originally
establishing nonproprietary names for the Neoral products in
1995.
To evaluate Novartis’s arguments regarding the FDA’s
initial assignment of nonproprietary names, we begin by
reviewing the statutory and regulatory framework governing
established names. There are two key statutory provisions.
The first is 21 U.S.C. § 352(e)(1)(A)(i), which provides that a
drug may be subject to charges of misbranding unless its label
bears, among other things, “the established name (as defined
in subparagraph (3)) of the drug, if there is such a name.”
Subparagraph (3) provides that “established name” means:
(A) the applicable official name designated pursuant to
section 358 of this title, or
13
(B), if there is no such name and such drug, or such
ingredient, is an article recognized in an official
compendium, then the official title thereof in such
compendium, or
(C) if neither clause (A) nor clause (B) of this
subparagraph applies, then the common or usual name,
if any, of such drug or of such ingredient [with an
exception that is not relevant here].
21 U.S.C. § 352(e)(3). Neither party argues that clause (C)
applies to the name given Novartis’s products in 1995, and
they affirmatively agree that clause (B) didn’t apply to those
designations. See Mem. Op. (Sept. 9, 2004) 28 n.12. The
“official compendium” referred to in clause (B) is the United
States Pharmacopoeia (“USP”); the applicable article (or
monograph) titles in the USP omit the term “microemulsion.”
Reasoning by process of elimination, Novartis argues that
with clauses (B) and (C) out of the picture the FDA’s 1995
action must have been a designation under clause (A) of
§ 352(e)(3). Br. for Appellants 34. But this hardly appears to
be the only reasonable reading of the statute. For starters,
§ 352(e)(1)(A)(i) refers to an established name “if there is
such a name,” suggesting that the categories of established
names set out in § 352(e)(3) do not exhaust the categories of
nonproprietary names that the FDA might assign.2
2
The parties appear to assume that § 352(e)(1)(A)(i)’s
labeling requirements encompass any nonproprietary name assigned
to a drug by the FDA even if the name does not qualify under
§ 352(e)(3). We express no opinion on the issue.
14
Novartis tries to buttress its reading by asserting that
§ 352(e)(3) defines the three name categories in “descending
order of preference.” Br. for Appellants 34. Novartis seems
to have no basis for the claim; every list puts some things
lower than others, but order is not necessarily an indication of
rank. Also, the FDA’s regulations take quite a different
position, providing that the FDA “will not routinely designate
official names under section 508 of the act [21 U.S.C. § 358].
. . .” 21 C.F.R. § 299.4(e).
The second key statutory provision is § 358. Subsection
(a) provides:
The Secretary may designate an official name for any
drug or device if he determines that such action is
necessary or desirable in the interest of usefulness and
simplicity. Any official name designated under this
section for any drug or device shall be the only official
name of that drug or device used in any official
compendium published after such name has been
prescribed or for any other purpose of this chapter.
21 U.S.C. § 358(a). Section 358(b) requires the Secretary to
undertake an apparently comprehensive review of the names
by which drugs are identified in official compendia—i.e., in
the USP. Section 358(c) provides that if, after such a review,
the Secretary determines that any names in the USP are
problematic in any of several specified ways, the Secretary
will initiate a notice-and-comment rulemaking to replace them
with new names.
Novartis’s argument that the FDA in fact designated an
official name pursuant to § 358 runs into trouble because the
FDA’s 1995 actions didn’t align with the requirements of
15
§ 358. First, there is no indication that the comprehensive
review described in § 358(b), at least arguably a prerequisite
to a § 358 designation, ever occurred. Second, as the district
court pointed out, § 358(c) requires the agency to designate
the official names through notice-and-comment rulemaking,
which wasn’t done in 1995. Novartis argues that the notice-
and-comment requirement was satisfied because the FDA
designated names for Novartis’s Neoral products “at the
conclusion of a formal process involving consultation with the
USP [Nomenclature Committee] and rejection of the existing
USP compendial name,” Br. for Appellant 34, but it never
explains why those procedures should be considered
interchangeable with notice-and-comment rulemaking. Nor
did the FDA publicly indicate in some other way in 1995 that
it was designating an official name under § 358. Third,
§ 358(a) requires that once the FDA designates an official
name, that name must be used in any official compendium.
But, as explained above, the USP Nomenclature Committee
did not adopt the nonproprietary names that the FDA assigned
to Neoral in 1995. Novartis disregards this statutory
requirement. Accordingly, there is no reason to believe such a
designation occurred. Cf. Mem. Op. (Sept. 9, 2004) 28-29.
Finally, Novartis contends that whether the FDA assigned
official names pursuant to § 358 is a factual question that
cannot be answered without supplementing the administrative
record with documentation of the FDA’s actions when it
originally approved nonproprietary names for the Neoral
products. We see no need for such an excavation. Given the
gaps identified above, there is no serious likelihood that extra
documentation of the process by which the FDA developed
the nonproprietary names it initially assigned to the Neoral
would overturn our conclusion that the FDA did not proceed
under § 358.
16
Having rejected Novartis’s characterization of the FDA’s
actions in 1995, we examine the agency’s explanation of the
names it then assigned to the Neoral products as “interim
established names.” The FDA argues that the reference in
§ 352(e)(1)(A)(i) to a drug’s established name, “if there is
such a name,” means that it was not limited to the options set
out in § 352(e)(3) in designating a nonproprietary name. It
explains that it had filled that statutory gap by creating a
fourth category of names for drug products: “interim
established names.” Recall that under § 352(e)(3) a drug
product’s established name could be the official monograph
title for that product in the USP. The USP Nomenclature
Committee acts on its own schedule, so that its designation of
a name qualifying under § 352(e)(3)(B) need not coincide
with the FDA’s approval of a drug. Mem. Op. (Sept. 9, 2004)
29-30. Given the variable sequence, the FDA’s designation of
“interim” or “provisional” established names outside the
§ 352(e)(3) triad appears both consistent with the statutory
structure and reasonable.
As the statute leaves space for FDA designation of
interim or provisional established names, Novartis has no
basis for claiming that § 358 commands a notice-and-
comment rulemaking for the change of such a name. And
Novartis makes no claim that the APA itself commands use of
that specific procedure in the absence of a § 358 designation.
Insofar as Novartis makes a general claim that FDA acts
arbitrarily and capriciously by not providing adequate notice
and opportunity for comment before making any change of
the Neoral products’ interim names, our earlier observation
controls: Novartis had ample notice and opportunity to
comment in its own citizen petition proceeding.
17
* * *
The judgment of the district court is
Affirmed.