Natural Resources Defense Council, Inc. et al v. United States Food and Drug Administration et al

Filing 51

DECLARATION of Jennifer A. Sorenson in Support re: 19 MOTION for Summary Judgment.. Document filed by Center For Science In The Public Interest, Food Animal Concerns Trust, Natural Resources Defense Council, Inc., Public Citizen, Inc., Union Of Concerned Scientists, Inc.. (Attachments: # 1 Exhibit A)(Sorenson, Jennifer)

EXHIBIT A
TO SUPPLEMENTAL DECLARATION
OF JENNIFER A. SORENSON
Nitrofurans, 56 Fed. Reg. 41,902 (Aug. 23, 1991)

41902

Federal Register / Vol. 56, No. 164 / Friday, August 23, 1991 / Rules and Regulations

DEPARTMENT OF HEALTH AND
HUMAN SERVICES
Food and Drug Administration
21 CFR Parts 510 and 558
[Docket Nos. 76N-0172 and 76N-0232]
Nitrofurans; Withdrawal of Approval of
New Animal Drug Applications
AGENCY. Food and Drug Administration,
HHS.
ACTION: Final rule; final decision
following a formal evidentiary public
hearing.
SUMMARY: The Commissioner of Food
and Drugs is issuing his final decision on
the proposal to withdraw approval of
the new animal drug applications
(NADAs) for two nitrofuran animal
drugs: furazolidone (NADAs 11-698, 9073, 12-061, 9-393, 13-805) and
nitrofurazone (NADAs 6-395, 8-142, 9415, 8-989, 10-741). The drugs are
labeled and approved for antiprotozoal
use for a wide variety of conditions in
poultry and swine.
The Commissioner has determined
that nitrofurazone and furazolidone are
not shown to be safe under the
conditions of use for which they were
approved under 21 U.S.C. 360b(e)(1)(B). 1
Additionally, the Commissioner finds
that furazolidone and its metabolites
have by substantial new evidence been
shown to induce cancer in man or
animals within the meaning of 21 U.S.C.
360b(d)(1)(I). Thus, he is withdrawing
approval for the drugs and is revoking
the regulations codifying the approval of
these applications in 21 CFR 510.515,
558.4, 558.15, and 558.262, and 558.370.
Also, he is affirming with modifications
the initial decision of the Administrative
Law Judge, who made similar findings.
EFFECTIVE DATE: September 23, 1991.
ADDRESSES: The transcript of the
hearing, evidence submitted, and all
other documents cited in this decision
may be seen in the Dockets
Management Branch (HFA-305), Food
and Drug Administration, rm. 1-23,
12420 Parklawn Dr., Rockville, MD
20857, from 9 a.m. to 4 p.m., Monday
through Friday.
FOR FURTHER INFORMATION CONTACT.
Robert L. Spencer, Division of
Regulations Policy (HFC-220), Food and
ISection 360b(e)(1)(B) contains a reference to
"subparagraph (H) of paragraph (1) of subsection (d)
* * "."
Because, inPub. L. 100-470, Congress

redesignated subparagraph (H) subparagraph (1),
as
the reference should read "subparagraph (I) of
°
paragraph (1) of subsection (d) * *."For purposes
of this final decision, FDA is interpreting the act as
if Congress had made this necessary conforming
change.

Drug Administration, 5600 Fishers Lane,
Rockville, MD 20857, 301-443-3480.
SUPPLEMENTARY INFORMATION: The

purpose of this proceeding is to
determine whether the Food and Drug
Administration (FDA) should withdraw
approval of the NADAs for use in foodproducing animals. The effect of this
decision is that these two drugs may no
longer be marketed in the United States,
nor may they be exported except as
allowed by law.
I. Introduction
The history of this hearing is set forth
in the initial decision (I.D.) and in the
notice of hearing (49 FR 34965,

September 4, 1984). That entire history
will not be repeated here. Briefly, this
consolidated proceeding involves two
animal drugs that have been used in this
country since the 1940's, in the case of
one of the drugs, and since the 1950's, in
the case of the other drug. The two
drugs, furazolidone and nitrofurazone,
are part of a chemical class referred to
as nitrofurans. In the 1960's, evidence
first surfaced that furazolidone caused
tumors in laboratory animals. As
evidence began to mount, FDA issued a
notice of opportunity for hearing on
March 31, 1971 (36 FR 5927), proposing
to withdraw the NADAs for
nitrofurazone on the grounds that it was
no longer shown to be safe. A similar
notice for furazolidone was issued on
August 4, 1971 (36 FR 14343).
Since that time, the sponsors of these
drugs (Hess and Clark and SmithKline,
sponsors) have brought new data before
the agency, which has reviewed the
data. A full evidentiary hearing has
been held to determine whether the
NADAs of these two drugs should be
withdrawn on the grounds that the drugs
are no longer shown to be safe, and, in
the case of furazolidone, whether its
NADA should be withdrawn under the
Delaney anticancer clause as well.
The Administrative Law Judge (ALl)'
issued his I.D. on November 12, 1986,
finding that the NADAs should be
withdrawn. The ALI found that
furazolidone was an animal carcinogen
that should be withdrawn under both
the Delaney clause (21 U.S.C.
360b(d)(1)(I), as incorporated in 21
U.S.C. 360b(e)(1)(B)) and the general
safety clause (21 U.S.C. 360b(e)(1)(B)).

He also found that nitrofurazone,
including its metabolites, is an animal
tumorigen, and, therefore, a suspect
carcinogepi that should be withdrawn
under the general safety clause. The ALJ
also found that the sponsors had failed
to provide a reliable method of residue
detection for either drug and that the
residues of neither drug have been

shown to be safe. In addition, he
determined that the concentrations of
residues of furazolidone were not shown
to be below the level of carcinogenic or
toxicological concern.
Since the issuance of the I.D., the
sponsors have filed briefs and
exceptions totalling over 350 pages that
take exception to virtually every
ultimate and supporting conclusion of
the AL, and that raise several legal and
procedural exceptions as well. 2
Following the filing of exceptions, on
August 25, 1987, the Center for
Veterinary Medicine (Center) moved to
reopen the evidentiary record in order to
receive National Toxicology Program
(NTP) draft reports of bioassays
involving nitrofurazone, one of the drugs
at issue here, and nitrofurantoin,
another nitrofuran but one not directly
at issue heree. See GF-1700. On
September 21, 1987, the two sponsors of
the NADAs also filed motions
requesting that these materials be
admitted in the record, and in addition
requesting that the case be remanded to
the ALJ for further testimony regarding
the issues raised by the NTP reports.
By an order dated November 2, 1987,
then Commissioner Frank Young

granted the motions by all parties to
reopen the record to admit the draft NTP
reports. In response to the sponsors'
motion to remand the matter for further
testimony, Dr. Young permitted a limited
remand to the ALJ. Under the terms of
the remand, each party was allowed to
submit written testimony concerning the
NTP reports from one expert Witness
who had already testified in the
proceeding. The remand order also
allowed I day of cross-examination to
be conducted before the ALJ. Finally,
the order allowed each party to submit a
supplemental brief following the hearing
on the NTP reports. Each party filed its
expert's supplemental testimony on
January 6, 1988. The hearing on remand
was held on February 3, 1988, and
supplemental briefs were filed on March
8, 1988. Since that time, the record in
this hearing has been officially closed.
After fully reviewing the evidence in
the administrative record and the
exceptions to the I.D. raised by the
sponsors, I find that there is clearly
enough evidence in the record to justify
the ALJ's conclusion that furazolidone
and nitrofurazone are no longer shown
to be safe.
2 The exceptions filed by the sponsors in this

proceeding.exceeded involume those filed in any
other hearing before FDA. Many exceptions were

frivolous or trivial.
s The final version of this report has been

published, but it does not differ from the draft as to

any conclusions pertinent to this hearing.

Federal Register / Vol. 56, No. 164 / Friday, August 23, 1991 / Rules and Regulations

41903

II

I also find overwhelming evidence in
the record to support the ALI's
conclusion that the sponsors have failed
to provide a reliable means for detecting
residues of these drugs and their
breakdown products in animal tissue.
Such a detection method is necessary to
enable FDA to ensure that no dangerous
residues enter the human food supply.
On the basis of the administrative
record, I find that I am unable to ensure
that foods derived from animals treated
with these drugs will contain no more
than safe levels of residues of
furazolidone, nitrofurazone, and their
breakdown products (metabolites).
Therefore, I am by this notice
withdrawing all NADAs for
furazolidone and nitrofurazone.
In doing so, pursuant to 21 CFR
12.130(d), I am adopting the I.D. as
issued with some modifications as
stated below. As to exceptions filed by
the parties, I am herein addressing only
those that I consider significant. I am
not required by law or regulation to
address every exception made-only
those raising "significant" issues.
Simpson v. Young, 854 F.2d 1429, 1434
(D.C. Cir., 1988); 21 CFR 12.120(b) and
12.130(c). Where I do not specifically
address an exception of Hess and Clark
(H&C) or SmithKline (SK), their
exceptions are overruled for reasons
stated in the Center's Reply to
Exceptions.
I am expressly not ruling on any
exception filed by the Center because I
believe that doing so is not essential to a
decision on the issues in this proceeding.
As a result, my failure to address a
particular exception by the tenter
should not be construed as either an
affirmance or an overruling of that
exception.
II. Initial Findings
1. I reaffirm the statement of the
allocation and formulation of the burden
of proof in the Commissioner's
diethylstilbestrol (DES) decision (44 FR
54852), September 21, 1979) and apply

that to this proceeding. Under both the
Delaney and general safety clauses,
approval may be withdrawn if "new
evidence," evaluated together with
previously existing evidence, shows that
the drug is not shown to be safe. "New
evidence" includes any evidence not
available at the time the application was
approved, tests by new methods, and
tests by methods not originally
considered applicable. There does not
appear to be an issue about the
"newness" of the evidence upon which
the Center relies. The evidence
concerning the nitrofurans was not
available at the time they were
originally approved.

The proponent of withdrawal, the
Center, has the burden of making the
first showing (i.e., that the drug is no
longer shown to be safe). Hess and
Clark, Division of Rhodia, Inc. v. Food
and Drug Administration,495 F.2d 975,
992 (D.C. Cir. 1974).4 In Hess and Clark
, the court found that FDA has "an
initial burden of coming forward with
some evidence of the relationship
between the residue and safety to
warrant shifting to the manufacturer the
burden of showing safety." Id. at 993. In
the Commissioner's DES decision,
Commissioner Kennedy adopted the
followifig formulation of the Center's
threshold burden:
[the Center] must provide a
reasonable basis from which serious

questions about the ultimate safety of DES
and the residues that may result from its use
may be inferred."
44 FR 54861.

Once the limited threshold burden has
been satisfied, of course, the burden
passes to the sponsors to demonstrate
safety. Id.
There does not appear to be a
significant difference between the
parties on the subject of the burden of
proof. In any case, I find that the ALJ
applied the correct standard.
2. I find that cost/benefit
considerations are irrelevant under both
the Delaney clause and the general
safety clause. I agree with the Center's
view that American Textiles
ManufacturersInstitute v. Donovan, 452
U.S. 490 (1981) is ample authority for the
proposition that clauses like the Federal
Food, Drug, and Cosmetic Act's (the act)
general safety clause do not permit,
much less invite, cost/benefit analysis.5
The sponsors do not seriously argue that
such an analysis would be applicable
where the Delaney clause applies.
3. The sponsors argue that the rodent
studies that indicted nitrofurans as
carcinogens did not satisfy good
laboratory practice (GLP) standards
and, thus, cannot satisfy even the
Center's limited threshold burden of
proof. I disagree. No one argues that
these studies were very good studies by
today's standards. However, despite
their faults, as explained below, the
4There are two Hess and Clark cases: Hess and
Clark, Division ofRhodia, Inc. v. Food and Drug
Administratian,495 F.2d 975 (D.C. Cir. 1974)
(hereafter Hess and Clark I); and Rhone-Poulenc,
Inc., Hess and Clark Division v. Foodand Drug
Administration,636 F.2d 750 (D.C. Cir. 1980]
(hereafter, Hess and Clark !M).
5In the Commissioner's DES decision, 44 FR at
54883, FDA said: 'The law is clear that FDA may
not consider socio-economic benefits In the
determination of the safety to human beings of a
new animal drug, and I am not prepared to conclude
that it permits consideration of human health
benefits."

data that they generated constitute
substantial evidence of
carcinogenicity-evidence which is
sufficient to satisfy the Center's
threshold burden.
. I should note that FDA's GLP
regulations were not even proposed
until several years after the nitrofuran
bioassays were completed. Even more
important, by the terms of the preamble
to the GLP regulations, "valid data and
information in an otherwise
unacceptable study which are adverse
to the product * * * may serve as the

basis for regulatory action. This
disparity in treatment merely reflects
the fact that a technically bad study can
never establish the absence of a safety
risk but may establish the presence of a
previously unsuspected hazard."
(November 19, 1976, 41 FR 51206 and
51212). To the same effect, see FDA's
similar statement in the preamble to the
final rule (43 FR 59990).
The report of the NTP ad hoc panel on
chemical carcinogenesis testing and
evaluation (HF-104) cannot be cited to
the contrary: "All studies must serve as
an adequate basis for regulatory
decisions even though they have
protocol deficiencies in number of
animals per group, number of dose
levels, absent clinical observations,
etc." HF-104, 12-4. The panel added that
"our intent is not to imply that previous
studies would or should be judged
inadequate on the basis of modern
criteria [emphasis added]." Id. at 13.
4. I need not and do not address the
question of whether hormonally
mediated carcinogens are subject to the
Delaney clause. This is because the
sponsors have not proven that any
compound that is the subject of this
hearing is a hormonally mediated
carcinogen. See, e.g., Denial of Petition
for Listing of FD&C Red No. 3 (February
1, 1990, 55 FR 3520, 3537, and 3541). See
also infra, pp. 37 ff. In addition, as
discussed elsewhere (i.e., see pp. 48 ff), I
find that none of the compounds that are
the subject of this hearing has been
shown to be safe within the meaning of
the general safety clause. 21 U.S.C.
360b(e)(1)(B).
5. I agree with the Center (main brief
at 82, n. 67) that 10 -6 is an appropriate
risk standard by which to judge
nitrofurans and their metabolites. The
sponsors, while not directly attacking
this standard, did suggest that FDA has
in the past allowed greater levels of risk,
but they have cited no FDA-approved
new animal drug for which higher levels
of risk from residue were found.

41904

Federal Register / Vol. 56, No. 164 / Friday, August 23, 1991 / Rules and Regulations

III. "New Evidence That Furazolidone
Causes Cancer in Man or Animals"
I will proceed now to consider in
some detail the adequacy of the Center's
"new evidence that furazolidone causes
cancer in man or animals."
A. Evidence of Carcinogenicity-The
FourNorwich Studies
The Center's new evidence that
furazolidone causes cancer consists of
four animal bioassays performed under
the auspices of Norwich-Eaton, the
original furazolidone NADA sponsor, in
1973 and 1974. GF-195a, GF-195b, GF196, and GF-197 (collectively referred to
as "the Norwich studies"). These studies
are summarized in the I.D. at pp. 19-23.
In addition to the Norwich studies, the
Center relies on mutagenicity studies to
demonstrate that furazolidone is a
mutagen. If furazolidone is
demonstrated to be a mutagen, that fact
would lend support to the contention
that furazolidone is a carcinogen.
The sponsors contend that the
Norwich bioassays are not reliable
indicators of cancer for a host of
reasons. The most important
deficiencies cited by the sponsors
include the allegation that the maximum
tolerated dose (MTD) was exceeded in
several of the tests, so that tumors
attributed to the carcinogenic affect of
furazolidone were, in fact, the result of
toxic stress. The sponsors also contend
that the incidence of neoplasms in
treated test animals was not statistically
significant or was within the historical
range for spontaneous tumor generation
in the test animals. The sponsors further
argue that positive indications of
carcinogenicity were based on improper
groupings of benign and malignant
tumors, or of different tumor types. The
sponsors also fault the Norwich studies
for failing to comply with GLP
regulations that were adopted by FDA
after these studies were completed.
Among the GLP deficiencies cited by the
sponsors were illness in the test animals
or impurities in the test substance,
which should invalidate the results of
the Swiss Mouse Study, according to the
sponsors.
To the extent that the Norwich studies
do indicate that furazolidone causes
benign or malignant tumors, the
sponsors argue that furazolidone does
not act as a "direct" carcinogen. Rather,
they contend, the evidence
demonstrates that furazolidone causes
cancer only in doses high enough to
distort hormone levels in the test
animals. It is the change in hormone
levels, the argument runs, that actually
"causes" cancer in the test animals. The
sponsors also claim that the Norwich

test data demonstrate that, at low
enough levels, the ingestion of
furazolidone will have no carcinogenic
effect. The sponsors also claim that,
because humans and rodents have
different hormones, it is unlikely that
ingestion of furazolidone-treated
animals could cause cancer in humans.
Regarding the mutagenicity tests, the
sponsors' strongest argument is that
furazolidone was only weakly
mutagenic or was shown to be
mutagenic only under conditions that
are unlikely to be duplicated in
mammals. Thus, they argue, these
mutagenicity studies are not a reliable
indicator of furazolidone's carcinogenic
potential.
After a thorough review of the
evidence and the arguments in the
record, I find, for the reasons stated
below, that the Norwich bioassays,
while imperfect, satisfy the Center's
initial burden of adducing new evidence
raising questions about the safety and
carcinogenicity of furazolidone that are
sufficiently serious to require the
manufacturers to demonstrate
furazolidone's safety.
I also find that the mutagenicity tests,
when considered together with the
Norwich studies, add further evidence
that furazolidone is, at the very least, a
suspect carcinogen, and at worst, is a
proven animal carcinogen. I also find
that the Norwich studies and the
mutagenicity tests, considered together,
are inconsistent with the sponsors'
claimsof a hormonal theory of cancer
induction.
1. Maximum Tolerated Dose
I agree with the sponsors that the
MTD was exceeded in certain dosage
groups of two of the studies.
Specifically, I find that the MTD was
exceeded in the high- and mid-dose
groups in the Sprague-Dawley High
Dose Study (GF-195b] and in the highdose group in the Fischer 344 Rat Study
(GF-196). HF-310, p. 21; HF-309, p. 9;'
GF-1617.1, pp. 9-10; GF-1623.1, p. 21a.
The MTD may also have been exceeded
in the mid-dose group in the Fischer
study (GF-196). HF-309, p. 9; HF-310, p.
21; GF-1617.1, pp. 9-10; Transcript
("Tr.") III, pp. 39, 45-6, 50.

However, in the low-dose SpragueDawley Study (GF-195a), I find that the
MTD was not exceeded in any test
group. HF-310, p. 14; GF-1617.1. p. 9. The
sponsors do not contend otherwise. As
to the Swiss Mouse Study, the fact that.
there were no early deaths in males 'is
evidence that the MTD was not
exceeded in males.G-1617.1, p. 12. The"
MTD may have been exceeded in
'
females. However, the weight gain noted
in treated animals was comparable to

that noted in control animals, suggesting
that the toxicity was not due to
overdosing. G-1617.1, p. 12; GF-1623.1,
p. 22. Even if the MTD was exceeded in
the mid- and high-dose females, the
results wouldjust confirm the effect
seen in lower doses. The results in these
mid- and high-dose animals, although
not demonstrating relevant
carcinogenicity, will not have shown
safety either. GF-1623.1. pp. 21-2.
Moreover, neither SK nor H&C argues
that the MTD was exceeded in the lowdose group of test animals in either the
High-Dose Sprague-Dawley Study (GF195b) or the Fischer Rat Study (GF-196).
I agree that the MTD was not exceeded.
based on evidence in the record
demonstrating that the test animals in
the low-dose groups in both the HighDose Sprague-Dawley Study and the
Fischer Rat Study did not suffer a
weight decrement exceeding 10 percent
and did not exhibit other characteristics
usually associated with toxic dosing.
GF-1623.1; Bryan, Tr. XII-67-8; GF1617.1, pp..9-10.
After reviewing the evidence
concerning every group of test animals
whose dosage did not exceed the MTD. I
find that, in every case, the animals
dosed with furazolidone developed
neoplasms that exceeded the controls'
rate of neoplasms, and that the
difference was statistically significant in
most cases.
Specifically, I find that mammary
tumors in female rats in the Low-Dose
Sprague-Dawley Study (GF-195a]
exhibited a statistically significant dose
response that is indicative of the
carcinogenicity of furazolidone. GF1615.1, p. 11.A also find that, in the Swiss
Mouse Study (GF-197), statistically
significant dose-response trends were exhibited respecting bronchial
adenocarcinomas or adenomas in both
sexes and for lymphosarcomas in males.
GF-1613.1, p. 8; GF-1615.1, p. 10.
In the Fischer Rat Study, I find that
the incidence of mammary tumors
exhibited by rats in the low-dose group
was statistically significant when
compared to the controls. GF-1617.1, p.
10. I also find that the low-dose Fischer
rats exhibited not only increases in
mammary tumors but also debreased
onset time, increased multiplicity and
increased malignancy, all of which
indicate that furazolidone is a
carcinogen at doses below the MTD.CF-1617.1, pp. 9-10; GF-1623.1, pp. 21-2.
In the HighrDose Sprague-Dawley
Study (GF-195b), I find that, even in the
low-dose group, whose dose did not
exceed the MTD, the evidence
demonstrates that 41, out of the 50
treated rats developed mammary

Federal Register / Vol. 56, No. 164 / Friday, August 23, 1991 / Rules and Regulations
tumors, while only 29 out of 50 control
rats developed mammary tumors. GF195b, p. 32; GF-1623.1, p. 22. Where so
large a number of low-dose females
developed mammary neoplasms in
comparison with the controls, I doubt
that acute toxic stress, rather than
furazolidone, is the cause. The toxic
stress argument is also inconsistent with
the clear dose-response relationships
generated by this study. GF-1623.1, pp.
11-12; GF-1612.1, pp. 6-7, 10; GF-1617.1,
pp. 6, 9, 11; HF-309, p. 16; Tr. X, p. 93; Tr.
IV, p. 153.
The fact that test animals in the lowdose groups in the Norwich studies
developed neoplasms at rates higher
than the controls did demonstrate that
findings of carcinogenicity in these
studies cannot be dismissed as a
byproduct of overdosing. In addition, the
types of tumors and neoplasms
developed by rodents in groups where
the MTD was exceeded do not differ in
type or locus from those found in groups
where the MTD was not exceeded. GF1617.1, pp. 9-12; GF-1623.1, pp. 21-2.
This continuity of tumor type across
dosage groups suggests that not all the
neoplasms observed in animals whose
doses exceeded the MTD can be
attributed to acute toxic stress. See GF1617.1, p. 11. While I would not rely
solely on test data from dosage groups
where the MTD was exceeded, I find
that the similarity of tumor types
between dosage groups above and
below the MTD provides additional
support for the finding that furazolidone
itself, rather than any overdosing,
caused neoplasms in the test animals
that are indicative of carcinogenicity.
2. Statistical and Biological Significance
The sponsors challenge findings in the
I.D. that the incidence of neoplasms in
treated test animals are statistically and
biologically significant. Statistical
significance is concerned with the
probability that a given test result
occurred by chance, rather than because
of the effect that the test is designed to
study. Biological significance is
concerned with whether the animal
harboring a lesion will ultimately
become diseased as a result of the
lesion. GF-1612.1, p. 2.
The ALJ found that statistical analysis
of the tumor data from the four Norwich
studies was insufficient to evaluate the
effects of furazolidone, and that an
evaluation of their biological
significance was necessary. I.D., p. 42.
The ALJ found the Norwich data to
provide ample evidence of biological
significance. I.D., pp. 42-6. The sponsors
challenge findings of biological
significance, arguing that mammary
tumors occur spontaneously at a high

rate in Sprague-Dawley and Fischer 344
rats (HF-309, pp. 5, 22; HF-310, pp. 15,
18, 26; Tr. III, pp. 57-8). The sponsors
also assert that important factors that
can affect the incidence, multiplicity,
and onset time of mammary tumorssuch as age, diet, environment, physical
stress, hormonal status, and
immunologic competence-were not
adequately controlled in the Norwich
studies. The sponsors further assert that
the mammary tumors found in treated
test animals were in fact the result of
hormonal disruption and generalized
physiological stress in aging animals
caused by toxic doses of furazolidone
that far exceeded the MTD. HF-309, pp.
22-3; HF-310, pp. 3, 18, 22.
For the reasons stated below, I find
that the incidence of neoplasms in test
groups whose dosage did not exceed the
MTD was, for the most part, statistically
significant. Since toxic stress cannot
explain away these tumors, which were
the same types of tumors found in the
higher dose groups, I find that the
Norwich bioassays provide ample
evidence that furazolidone is an animal
carcinogen. Moreover, the increased
multiplicity of tumors, decreased onset
time, and increased malignancy of
tumors in all groups of test animals fed
furazolidone are additional evidence
that the tumor findings generated by
these studies are biologically
significant-i.e., that the findings are
indicative of the actual or potential
carcinogenicity of furazolidone. See p.
20, supra.
While I agree with the sponsors that
age, hormonal status, physical stress
and immunologic competence may have
some effect on cancer rate, I am
concerned that these factors cannot be
controlled in either the target animal
population that is fed furazolidone or in
the human population that eats food
products derived from these animals.
Therefore, I reject the sponsors'
invitation to ignore test findings raising
safety questions where these factors
were not controlled.
Accordingly, where, as here, four
different animal bioassays involving
two different species of rat and one
species of mouse all demonstrate that
treated test animals have an increased
rate of neoplasms even at doses below
the MTD, I find this to be biologically
significant evidence that the test
substance is an animal carcinogen. The
bioassays are treated individually
below.
a. The Low-Dose Sprague-Dawley.
Study.-Regarding the Low-Dose
Sprague-Dawley Rat Study (GF-195a),
the sponsors assert that the incidence of
mammary tumors in treated females .

41905

was not statistically significant. G-195a,
p. 9; GF-1631.1, p. 9; GF-1616.1, p. 11;
HF-310, p. 28. However, the sponsors
failed to consider time-to-tumor
information or to adjust for differential
mortality among dose groups. GF-1623.1,
pp. 10-11; GF-1612.1, p. 10; GF-195a, p.
6; HF-310, p. 28; HF-309, p. 16; GF1617.1, p. 9; GF-1615.1, p. 11; GF-1280, p.
17. Proper statistical analyses of tumor
data adjust for different mortality
among dose groups. See HF-104, pp.
210-14. Also, the sponsors failed to test
for dose-response trends, which make
more efficient use of the data and are
generally more sensitive in detecting
effects than are individual comparisons
of each dosage group with the control
group. GF-1613.1, p. 2; HF-104, pp. 20910.
In reviewing the results of the LowDose Sprague-Dawley Rat Study, I find
a statistically significant increase in
mammary neoplasms in females with
increasing doses of furazolidone, with
P=0.006 when using a trend test and
incorporating corrections for differential
mortality among the dose groups. GF1615.1, p. 11; GF-1280, p. 17. I find that
the statistical analyses conducted by the
Center are valid and in accord with
analyses conducted by the NTP (HF104). 1 also find that the results in the
Low-Dose Sprague-Dawley Study are
biologically significant. In addition to
showing a statistically significant
increase in mammary tumors in dosed
females, the test results show increased
multiplicity of mammary tumors in
female rats as the dosage of
furazolidone increased. GF-195a, p. 6.
When the multiplicity is expressed as a
percentage, the rate is monotonic (i.e.,
goes in one direction only), ascending,
dose-related, and significant. GF-1623.1,
pp. 11-12; Tr. IV, p. 153.
A witness for the sponsors testified
that the NTP rejects multiplicity of
mammary neoplasms in rats as an
indication of carcinogenic potential. Tr.
XV, pp. 72-3; GF-195a, p. 56. 1 find that,
to the contrary, the NTP draft reports on
nitrofurazone (GF-1700, p. 11) and
nitrofurantoin (GF-1701, p. 7) list
"multiplicity in site-specific neoplasia"
as one of the several "key factors" to be
considered when evaluating bioassay
test data for findings of carcinogenicity.
The same witness observed that the
incidence of rats in the study with single
mammary tumors went down as the
dosage of furazolidone increased. Tr.
XV, pp. 72-3; GF-195a, p. 56. This
statement is misleading. The test results
in the Low-Dose Sprague-Dawley study
demonstrate that the proportion of
animals with mammary tumors
increased with dose and that the

41906

Federal Register / Vol. 56, No. 164 / Friday, August 23, 1991 / Rules and Regulations

proportion of animals with multiple
mammary tumors Increased with dose.
GF-195a, pp. 6, 56; Tr. IX-47; IV-150-3.
Obviously, all that has happened is that
the proportion of animals in the study
with the more severe conditionmultiple mammary tumors-has
increased with dose, decreasing the
proportion of animals with the less
severe condition of only a single
mammary tumor.
In addition, Norwich, the original
study sponsor, conceded that two of the
three doses in the study significantly
increased tumor multiplicity and
"caused significantly earlier onset time
of mammary neoplasms and caused
significantly decreased survival rates
when compared to control female rats."
GF-195a, pp. 9-10, 50. The sponsors
assert that the decrease in mean time-topalpable-tumor was only marginally
significant in the mid- and high-dose
females and was not significant in the
low-dose group. However, I find that,
after adjusting for the differences in
tumor onset times between control and
treated animals, there was an increased
evidence of benign and malignant
mammary gland neoplasms in treated
females. GF-1623.1, pp. 10-11; GF1612.1, p. 10. These were biologically
significant. GF-1623.1, pp. 11-12; Tr. XII55-6; HF-104, p. 167. Also, I find that
when the decrease in onset time in the
mid-dose and high-dose groups is
considered in conjunction with the
statistically significant increases in
mammary tumors and with the doserelated increase in multiplicity, it
provides additional evidence of the
carcinogenicity of furazolidone. GF1612.1, p. 8; GF-1617.1, p. 5; GF-1623.1,
pp. 11-12; HF-104, pp. 167, 200-14; Tr.
IV, p. 153.
I also find that males in the mid-dose
and high-dose groups in the Low-Dose
Sprague-Dawley Rat Study exhibited an
increase in thyroid follicular adenomas
that increased with dose level. GF-195a,
p. 24. There is no evidence in the record
that a statistical analysis was conducted
on these data. Notwithstanding the lack
of statistical analysis, the dose-related
increase in thyroid follicular adenomas
in the mid- and high-dose males is still
noteworthy. The same tumor was found
in dosed males in the High-Dose
Sprague-Dawley Study (GF-195b, pp. 28,
36-64; GF-1623.1, p. 11; GF-1612.1, p. 10;
Tr. IX-135; Tr. X-41-2 and in the Fischer
Rat Study GF-196, pp. 4, 9-11, 26-7, 3464: GF-1823.1, p. 10; GF-1612.1, p. 11;
HF-309, v. 8: HF-310, pp. 21, 23). I find
that: (1) the increased incidence of
thyroid follicular adenomas in male rats
in three different studies; and (2) the
findings of mammary adenomas in

females in all four studies combine to
provide significant evidence that
furazolidone is an animal carcinogen.
b. The High-DoseSprague-Dowley
Rat Study. The sponsors' main attack on
this study is that the dosage levels
exceeded the MTD and that the tumors
seen in this study were the result of
acute toxic stress. However, although
the MTD was exceeded in the high- and
mid-dose groups, this finding does not
explain away the results generated by
this study.
First, I note that, in the low-dose
group alone, where the dose -did not
exceed the MTD, 41 out of the 50 treated
female rats developed mammary tumors,
while only 29 out of 50 female control
rats developed such tumors. GF-195b, p.
24. Unfortunately, I can find no evidence
in the record that this comparison was
analyzed for statistical significance.
However, when a statistical analysis
was performed using only the low- and
mid-level dose groups in this study, the
incidence of mammary tumors was
found to be statistically significant after
adjusting for differential mortality. GF1613.1. pp. 3, 4, 6, 9. Because the same
types of tumors were observed in the
mid-dose group as in the low-dose
group, it is clear that not all the tumors
in the mid-dose group can be explained
away as the result of overdosing. GF1617.1, pp. 6, 9; GF-1623.1, p. 11; GF-

1612.1, p. 10. Therefore, I find that the
statistical significance of the incidence
of mammary tumors in treated female
rats in the low- and mid-dose groups in
the High-Dose Sprague-Dawley Study is
evidence of the carcinogenic property of
furazolidone.
The evidence demonstrates a
statistically significant increase in
thyroid follicular adenomas in treated
male rats, with P=0.0003 when using a
trend test and incorporating corrections
for differential mortality among the dose
groups. GF-195b, pp. 28, 36-64; GF1615.1, p. 6; Tr. IX,p. 135; Tr. X, pp. 41-2.
Because this calculation includes dosage
groups that exceeded the MTD, I would
not base a finding of furazolidone's
carcinogenicity on this fact alone.
However, when this fact is considered
together with other relevant evidence in
the record, I find that it is further
evidence of the carcinogenic potential of
furazolidone. The fact that treated male
rats in all three of the Norwich studies
that used rats developed the identical
tumor, including rats in the Low-Dose
Sprague-Dawley Study, suggests that
this finding is not the result of
overdosing. GF-195a, p. 24; GF-195b, pp.
28, 36-64; GF-196, pp. 4, 9-11, 26-7, 3464: GF-1623.1, pp. 10-11.

The High-Dose Sprague-Dawley Study
contained much the same evidence of
biological significance as did the Fischer
Rat Study and the Low-Dose SpragueDawley Study. For example, the HighDose Sprague-Dawley showed a doserelated increase in multiplicity of
mammary tumors and a decreased onset
time in treated females. GF-195b, pp. 3,
8, 14-15, 26, 32-3, 36-64; GF-1623.1, p. 11,
GF-1617.1, p. 9; HF-309, p. 16. I find
substantial credible evidence in the
record that both of these factors are
biologically significant evidence of
carcinogenicity. GF-1623.1, pp. 11-12;
HF-104, pp. 167, 210-214; GF 1615.1, p. 4:
GF-1612.1, pp. 6-7; Tr. IV, p. 153; Tr. X,
p. 93.
In addition to this evidence, the data
also showed a statistically significant
increase in neural astrocytomas in
males, both in all dosage groups and in
just the two lower dosage groups, when
the data were adjusted for differential
mortality rates among the groups. GF195b, pp. 28, 36-44; GF-1623.1, p. 11; GF1612.1, p. 10; GF-1613.1, pp. 3-4, 6-9; HF309, p. 16; HF-310, pp. 19-20. While I
would not base a judgment of
furazolidone's carcinogenic potential on
this fact alone, I find that, when weighed
with the other evidence in the record,
the increased incidence of neural
astrocytomas in males is additional
evidence pointing to the ultimate finding
of carcinogenicity. Tr. IV-121; Tr. X-3638, 44.
When all of the above evidence is
considered, i.e.,-the dose-related.
statistically significant generation of the
tumors reported in this study; the large
increase in tumors in the low-dose
group, the additional factors evidencing
biological significance; and the
similarity of these findings with similar
studies, as a whole, the evidence from
this study is inconsistent with the
sponsors' assertions that the tumors
reported in this study were the result of
overdosing..
c. The FischerRat Study. In the
Fischer Rat Study (GF-196), as noted
earlier, even if we limit our review to the
low-dose group, which received a dose
of furazolidone that was below the
MTD, a statistically significant increase
in mammary neoplasms in treated
animals was demonstrated. GF-1617.1,
pp. 9-10.
The sponsors complain that benign
and malignant tumors should not have
been grouped together for the purposes
of analysis. While I disagree with the
sponsors for reasons that will be
detailed in a separate section, I note
that, even without combining benign and
malignant tumors, mammary
adenocarcinomas (malignant tumors)

Federal Register / Vol. 56, No. 164 / Friday, August 23, 1991 / Rules and Regulations
alone exhibited a statistically significant
dose-related increase in the three
dosage groups in this study. GF-1615.1,
p. 10. 1find that the two factors listed
above-the statistically significant
increase in mammary adenocarcinomas
in females in the low-dose group (which
were not dosed above the MTD, GF1617.1, p. 9) and the statistically
significant increase In malignant
mammary neoplasms in all dosage
groups--are biologically significant
evidence that furazolidone is an animal
carcinogen. GF-1617.1, pp. 6, 9-10.
In addition several other indicators of
furazolidone's carcinogenicity were
found in the Fischer Rat Study. When all
three dosage groups were considered,
test animals fed furazolidone exhibited
increases in mammary neoplasms with
decreased onset time, increased
multiplicity, and increased malignancy.
GF-1617.1, pp. 9-10; GF-1623.1, pp. 21-2.
While the sponsors complain that data
from the mid- and high-dose groups
should not be considered because the
dose exceeded the MTD, I find that the
continuity of tumor type as the dosage
increased allows us to consider these
findings as additional indications that
furazolidone is an animal carcinogen.
As noted earlier, I also find it
biologically significant that males in this
study developed the same type of
tumor-adrenal follicular adenomas-as
did the male rats in the Low-Dose
Sprague-Dawley Study (in which no
dosage group exceeded the MTD) and
the High-Dose Sprague-Dawley Study.
GF-1623.1, pp. 10, 14-15; GF-1617.1, p.
10; GF-1612.1, p. 11; HF-309, p. 8; HF310, pp. 21, 23; GF-196, pp. 4, 9-11, 26-7,
34-.64. Moreover, furazolidone
demonstrated a dose response as to
these tumors in this study. CF-1615.1, p.
9; GF-1280, p. 11; GF-1613.1, p. 9. I find
this to be additional evidence that
furazolidone is an animal carcinogen.
d. The Swiss Mouse Study. The
sponsors argue that the data in the
Swiss Mouse Study (GF-197) are not
biologically significant because, after
the treatment period ended, the midand high-dose females and the high-dose
males suffered a high mortality rate that
is indicative of servere toxic stress. The
sponsors argue that, whether this high
mortality was due to environmental
factors, intercurrent infection, or doses
exceeding the MTD, the study is too
flawed to proride evidence on the issue
of whether furazolidone causes lung
cancer.
I disagree. First, statistically
significant dose-response trends for
bronchial adenocarcinomas and/or
adenomas In both sexes and for
lymphosarcomas in males were
reported. GF-1013.1, p. 8; GF-1615.1, p.

10. If the tumors were produced by
environmental factors or from doses

exceeding the MTD, I would not expect
to find the clear dose-response
relationship that this study evidences. In
addition. I agree with the Center that the
Swiss Mouse Study may actually
understate the incidence of tumors
expected from t lifetime exposure to
furazolidone. GF-1623.1, pp. 23-4; GF1617.1. pp. 7--8. This understatement may
have occurred because test animals
should be exposed to the test substance
for 24 months in the standard bioassay
(HF-104, p. 188). In the Swiss Mouse
Study, by contrast, the test animals were
dosed for only 13 months (GF-197, p. 5;
HF-309, p. 19) but nevertheless
produced positive results. Thus, I find
that the data are at least as likely to
understate the carcinogenic effect of
furazolidone as they are to overstate it.
3. Combination of Tumor Type
The sponsors assert that benign
tumors should not be considered in
assessing the carcinogenicity of
furazolidone, and that benign tumors
should not be grouped together with
malignant tumors for the purpose of
statistical analysis. The sponsors also
complain that different types of skin
tumors were improperly grouped
together for the purposes of analysis.
Benign neoplasms are considered to
be indicative of cancer because benign
and malignant tumors often arise in the
same tissue and may represent a
spectrum of tumor development and
progression. GF-1623.1, pp. 13-14. In the
Fischer Study (GF-196) and in the LowDose and High-Dose Sprague-Dawley
studies (GF-196a and GF-196b,
respectively), benign and malignant
mammary tumors were grouped together
because benign mammary tumors can
progress to malignancy, because they
arise in common tissue (mammary
epithelium), and because of differences
in diagnosis from one pathologist to
another. GF-1623.1, pp. 13, 16; Tr. III, p.
84. 1find that the grouping of benign and
malignant mammary tumors was proper
in these circumstances.
I also note that, while the sponsors
rely on a finding of the International
Agency for Research on Cancer that
only malignant neoplasms provide
evidence of cancer (see HF-104, p. 279),
the NTP, an arm of the Department of
Health and Human Services that was
set up to conduct toxicology studies,
does consider the increase in benign
tumors and an increase in a combination
of benign and malignant tumors, under
appropriate conditions, when evaluating
carcinogenicity. HF-104, pp. 226-229,
232: GF-1700, p. 11; GF-1701, p. 7.

41907

I find that. based on the common
organ and tissue site and the known
tendency of mammary neoplasms to
progress to cancer, the consideration of
benign mammary neoplasms and their
combination with malignant mammary
tumors for the purpose of analysis were
appropriate in the Norwich studies. I
also find that there is no credible or
sufficient evidence in the record to the
effect that any known tumorigen causes
only benign tumors. I also find that,
because the decision to withdraw the
NADAs for furazolidone rests on the
general safety clause as well as the
Delaney clause, the evidence in the
record that furazolidone causes an
increased incidence of benign mammary
neoplasms in treated test animals which
received doses below the MTD is
evidence that, when considered in
conjunction with evidence of
mutagenicity, supports the conclusion
that furazolidone is no longer shown to
be safe.
I further find that the combination of
various types of skin tumors for the
purposes of analysis was proper to
determine that carcinogenicity or
tumorigenicity of furazolidone.
Combining skin carcinomas and
epitheliomas is acceptable under the
NTP guidelines (HF-104, p. 232). These
types of tumors gave statistically
significant dose-response relationships
in Fisher 344 rats. GF-1613.1, p. 8. While
I would not base a finding of
furazolidone's carcinogenicity or
tumorigenicity on skin tumor data alone,
I find that it is additional relevant
evidence that, when considered with the
other evidence in the record, helps
demonstrate the carcinogenic and
tumorigenic properties of furazolidone.
In summary, I find that the four
Norwich studies, taken as a whole.
provide enough evidence of
furazolidone's carcinogenic potential to
meet the Center's burden of
demonstrating new evidence raising
questions about the safety of
furazolidone that are sufficiently serious
to require the sponsors to demonstrate
furazolidone's safety, which they have
not done. In each of the four studies, the
tumor types were biologically significant
because each of them has the potential
to affect adversely the health of the
animal in which they were observed.
Moreover, feeding furazolidone to
rodents significantly increased the
incidence of each type of tumor, and,
where mammary neoplasms occurred, it
increased -theirmultiplicity and
decreased the time to tumor when
compared to rodents that were not fed
furazolidone. GF-1623.1, pp. 11-2.

41908

Federal Register / Vol. 56, No. 164 / Friday, August 23, 1991 / Rules and Regulations

4. Historical Range of Tumor
Development
The sponsors claim that the rates of
mammary, skin and thyroid tumors
observed in treated animals in the
rodent studies were within the range of
historical variation in spontaneous
incidence for these tumors. HF-310, p.
22; HF-309, pp. 22, 25. However, the
evidence of record does not support the
sponsors' claim. I find that the incidence
of mammary tumors in the control
female Fischer rats of 20 percent (10/49)
is below the historical range reported in
the record of 31 percent to 46 percent.
GF-1413.1, p. 1451; HF-257, p. 10. The
incidence of mammary tumors in the
low-dose group alone is 28/50, or 56
percent. GF-196, p. 26. I therefore find
that the incidence of mammary tumors
in treated females in the low-dose group
alone in the Fischer Rat Study exceeds
the historical range, providing additional
evidence of furazolidone's carcinogenic
properties.
The record also contains several
reasons why tumor incidence may vary
from study to study. HF-310, p. 22. This
is the reason why valid scientific test
protocols require that concurrent control
animals be compared with a test group
of treated subjects. This concept of
concurrently controlled studies is basic
to scientific investigation, and FDA
cannot allow historical data to
contradict concurrently controlled
studies.
5. Hormonal Induction
The sponsors argue that, to the extent
that furazolidone and nitrofurazone
cause tumors, they do so through a
hormonal mechanism which occurs only
at dose levels over a threshold and,
therefore, are not subject to the Delaney
clause because the threshold is above
any likely human consumption levels.
Based on the record, I draw three
scientific conclusions that militate
strongly against the argument that
furazolidone's tumorigenicity is based
solely or even primarily on a hormonal
mechanism. First, the increase in nonendocrine tumors discussed In GF1623.1, GF-1613.1, p. 8, and GF-1615.1, p.
10 is important in showing that.a
genotoxic (i.e., damaging to
deoxyribonucleic acid, thus causing
mutations or cancer) mechanism is
almost certainly responsible.
Second, the positive results of
mutagenicity tests on furazolidone
contradict the hypothesis that hormonal
induction is the sole mechanism by
which the substance induces cancer.
GF-709; GF-710; GF-829; GF-833; GF834; GF-849; GF-850; GF-1620.1, p. 9.

Third, the failure to demonstrate
increased plasma progesterone levels in
orally dosed animals means that the
target organs for carcinogenic action
were not exposed to increased
progesterone levels. GF-1018, table 8;
HF-310, pp. 4-11. Thus, the hormone
hypothesis is clearly refuted by the
sponsors' own data.
Against these facts, the sponsors cite
what they believe is evidence to the
contrary. I will consider their
contentions.
The sponsors contend that the LowDose Sprague-Dawley Rat Study (GF195a) demonstrates that furazolidone,
unlike direct acting carcinogens, causes
tumors only at dose levels that cause
hormonal disruption. HF-309, p. 29.
However, as stated above the rats in
this study did develop tumors,
demonstrating a dose response,
including tumors at doses below those
that would cause "hormonal disruption."
Thus, the sponsors' entire argument
about a hormonal mechanism based on
this study has a false premise.
The sponsors cite as "compelling
evidence" supporting their hormonal
theory (H&C exceptions, p. 114) studies
showing that ovariectomy has been
shown essentially to eliminate the
occurrence of mammary tumors in
furazolidone-treated rats, while
*significant numbers of tumors occurred
in nonovariectomized rats.
However, ovariectomy of rats also
reduces the incidence of mammary
tumors induced by known carcinogens
such as 3-methylchloranthrene (3MC)
and N-nitrosomethylurea. GF-1417; GF1616.1, p. 12. Both of these compounds
are known to be potent genotoxic and
carcinogenic substances. GF-1616.1, p.
12. Ovariectomy also reduced the
control incidence of mammary tumors
from 20 percent to 0 percent in female
rats. GF-430, p. 13. Therefore, the
diminution of tumors after ovariectomy
is not.evidence of the absence of a
genotoxic mechanism.
The sponsors suggest that
furazolidone blocks the synthesis of
corticosterone, leading to enhanced
production of progesterone and other
corticosteroids, which in turn results in
mammary hyperplasia. HF-310, pp. 3-11.
This the sponsors consider to be further
evidence of the existence of a hormonal
mechanism.
On the contrary, a feeding study of the
effect of furazolidone on plasma steroid
levels, GF-1018, Table 8, showed that
there was no increase in the plasma
levels of progesterone at the highest
dosage level. Thus, the thesis that
increased progesterone levels caused by
furazolidone are responsible for
mammary tumors gains no support. The

sponsors attempt to explain away the
fact of decreased plasma progesterone
levels at the high furazolidone dose by
invoking a complex "adrenal adaption"
theory, but their "evidence"
acknowledges that "weather [adrenal
adaption] could lead to mammary tumor
formation remains obscure." GF-1011, p.
8. Hence, the sponsors have adduced no
evidence for this theory.
I find that the data support the
proposition that furazolidone can act as
a direct carcinogen: in intact rats, no
plasma progesterone increases were
seen (GF-1018, Table 8); no change in
progesterone-sensitive organs was seen
(GF-195b); and mammary tumors were
induced. GF-195b, pp. 32-3.
The sponsors also argue that the
patterns of tumorigenesis in the four
Norwich studies are "characteristic" of
hormonal disruption (SX-187, pp. 6-7:
Tr. IX-20A; HF-309, pp. 8-9), but their
theory fails to explain the statistically
significant increase in nonendocrine
tumors found in these studies. See
supra,pp. 19 and 30 and GF-1613.1, p. 8;
GF-1615, p. 10; GF-1623.
Further, the sponsors argue that the
hormonal mechanism in the rat is not
duplicated in human physiology because
the function of corticosterone in the rat
is performed by cortisol in humans.
Because of this difference, they say, the
hormonal derangements caused by
blocking the synthesis of corticosterone
in the rat is less likely to occur in
humans. Tr. X-63, 73. According to the
sponsors, the evidence shows the rat to
be a poor model for predicting the
effects of furazolidone in humans
because corticosterone is not the
primary messenger regulating human
hormonal balance. HF-309, pp. 3-4, 6-8,
15-9; HF-310, pp. 4-11, 27-30.
However, my examination of the
evidence has revealed that the hormonal
mechanism of tumor induction is not
unique to the rate but has a
physiological analog in man. Tr. X-6165; Tr. IV-108-111. Hence, the difference
between cortisol and corticosterone
does not constitute a reason why
furazolidone would not have a similar
effect in humans.
To conclude, whether or not hormonal
changes may occur as a result of acute
treatment with furazolidone, as argued
by the sponsors, such a mechanism
cannot be invoked as the only tumorinducing mechanism given the evidence
of the presence of (1) nonendocrine
tumors (GF-1613.1, p. 8, GF-1615.1, p. 10,
GF-1623). (2) mutagenic activity (GF849; GF-850), and (3) the failure of
furazolidone to elevate plasma
progesterone in any long-term feeding
study. GF-1011, pp. 7-8; GF-1018, p. 18.

Federal Register / Vol. 56, No. 164

/ Friday, August 23, 1991 / Rules and Regulations

In fact, the sponsors have not proven
that the tumors in the Norwich studies
were induced solely by hormonal
imbalance. Hence, I reject the sponsors'
argument that furazolidone tumors were
hormonally medicated.

(GF-1618.1, pp. 5, 7; GF-883; GF-884;
GF-1078, p. 39; GF-1007, p. 33). These
residue levels far exceed the 0.5 ppm
level claimed by the sponsors to be of
no carcinogenic concern. SX-182, p. 7;
HF-307, pp. 5-6; SX-183, p. 15; Tr. X-17-

B. Residue Detection
Having determined that furazolidone
is an animal carcinogen at worst, and a
tumorigen and suspected carcinogen at
best, I now must determine whether
residues of furazolidone would remain
in animal food products after
furazolidone had been given to the
animal under the current labeling
instructions and whether those residues
raise concerns about safety. This
determination is necessary under the
DES proviso to the Delaney clause (21
U.S.C. 360b(d)(1)(I(ii)) and is also
necessary under the general safety
clause. Section 360b(d)(2J(A) states that,
in assessing the safety of a drug, I must
consider "the probable consumption of
such drug, and of any substance formed
in or on food because of the use of such

19.

drug * *."
The sponsors have attempted to
demonstrate that, under the method of
analysis they have proposed, no
residues of furazolidone are found in
test animals that are 0.5 ppm or greater.
H&C exceptions at 132 ff. The sponsors
further assert that only furazolidone,
and not its metabolites, is covered by
the Delaney clause. The argument is
based on FDA's regulatory treatment of
other chemicals. SK exceptions at 30-33.
According to the sponsors, the phrase,
"such drug," as used inthe "DES
Proviso" to the Delaney clause, 21 U.S.C.
360b(d)(1)(I)). refers only to the new
animal drug which is the subject of the
NADA and which has been shown to
induce cancer under the Delaney clause.
The sponsors contend that the term,
"such drug," does not include the
metabolites or degradation products of
the drug and charge that the ALJ erred
in his interpretation of the Delaney
clause by stating, on pages 8, 9, and 13
of the LD. that the residue includes both
the parent drug and its metabolites. SK
exceptions at 30 ff. The sponsors further
argue that, to the extent the metabolites
of furazolidone are in question, the
metabolites are incapable of harming
consumers of food products that may
contain these metabolites. H&C
exceptions at 127 ff.
After reviewing the evidence and the
relevant portions of the statute, I must
disagree with the sponsors on every
point. First. I find credible ievidence in
the record that residues of furazolidoiae
as high as 3.62 ppm were recovered in
animals fed furazolidone under
conditions of use specified in the label

I also find that both the general safety
clause and the Delaney clause require
the agency to consider the effect that the
consumption of drug residues, including
metabolites, will have on human
consumers. As noted above, the general
safety clause, 21 U.S.C. 360b(d)(2)(A),
specifically requires the agency to
consider this factor when reviewing an
original application for an NADA. When
the agency considers whether to
withdraw an NADA for safety reasons
under section 360b(e](1) of the act, the
agency certainly may consider the
safety factors mandated by Congress in
section 360b(d). See DES
Commissioner's Decision, 44 FR 54852.
To hold otherwise would be inconsistent

with the clear intent of Congress in
passing safety legislation intended to
protect the American public from
ingesting potentially harmful drug
residues in food products.
These sponsors' arguments that
nitrofurans' metabolites are not of
carcinogenic concern are both contrary
to principles acknowledged by the
parties (Combined Critique of Center for
Veterinary Medicine's Allegations of
Facts, J 208-9) and the law of this
proceeding (49 FR 34971 and 34973,
September 4, 1984j.G
More importantly, interpreting the
Delaney clause so as not to defeat its
purpose requires that FDA find that the
clause comprehends metabolites as well
as parent drugs. The Center reminds us
(Replies to Exceptions, pp. 26-7) that
animal drugs may (1) be less
carcinogenic than their metabolites, (2)
leave no trace of parent compound in
the edible tissue of the treated animals,
and (3) cause no adverse effects to the
treated animals. Hence, the sponsors'
interpretation would compel FDA to
conclude that dangerous human
carcinogens could not be banned under
the Delaney Clause. I reject this
interpretation.
H&C claims that the court in Hess and
Clark laccepted its interpretation of the
term "'residue." However, the language
to which H&C refers, 495 F.2d at 991,

was, in context, a reference to H&C's
argument that the residues were actually
attributable to the impurity, "pseudoin the absence of information .to the

contrary. all drug-related residues including
metabolites are presumed to be potential
carcinogens, end must be taken Into account in
determining if there is 'no residue.' 49 FR 34973.

41909

DES," not DES residues themselves.
Neither is H&C's reliance on Scott v.
FDA, 728 F.2d 322 (6th Cir. 1984) apt.
There, the court found that a food
additive containing a carcinogenic
impurity is not subject to the Delaney
clause if the additive, when tested as a
whole, does not cause cancer. Here,
furazolidone and its metabolites have
been shown to cause cancer.
Alleged examples of FDA actions
contrary to this position do not form a
basis for a contrary conclusion. The
sponsors have cited no published FDA
document, much less a binding policy
statement, in which FDA concluded that
the Delaney clause does not apply to
metabolites. Nor have they cited a single
chemical regulated in a contrary
manner.
For the reasons stated above, I find
that the Delaney clause does apply to
carcinogenic'metabolite residues.
Therefore, it becomes clear that the
sponsors' proposed method of residue
detection fails to meet the standards
derived from the statute. The sponsors
concede that their chosen method of
residue detection--the Winterlin
method-does not measure total
residues, but only residues of the parent
compound. HF-260; SX-183, pp. 4-5, Tr.
X-11. The Winterlin method of analysis
would still be acceptable if the sponsors
had provided data demonstrating that
the depletion of the measured entity (the
"marker") from the measured animal
tissue (the "target tissue") bore a known
relationship to the depletion of all drug
residues of toxicological or carcinogenic
concern (December 31, 1987, 52 FR 49582
and 49583); GF-1610.1, p. 4. However,
the sponsors have failed to do so.
Hence, they have failed to adduce an
acceptable method of residue detection
that would permit FDA to determine
that furazolidone residues remaining in
treated animals would be safe to
consumers.
The sponsors claim that the evidence
demonstrates that none of the
metabolites of furazolidone remaining in
treated animals would be harmful to
consumers. SX-180, p. 3; SX-181, pp. 3-4;
SX-182, p. 4. For example, the sponsors
claim that the presence of the 5-nitro
group in nitrofuran compounds is
essential for any mutagenic or
carcinogenic activity resulting from its
partial reduction into reactive
intermediates. SX-182; SX-181; SX-182;
HF-308; SF-36.

However, my review shows that the
evidence indicates that metabolites of
furazolidone without the 5-nitro group
do have some mutagenic activity.
Aminofuran and acetamidofuran, for
example, tested both With and without

41910

Federal Register / VoL 56, No. 164

activation, are mutagenic. HF-97, Table
10. Thus, I find that nitroreduction does
not necessarily preclude subsequent
toxicity.
The evidence shows that there are a
number of different metabolic pathways
for the breakdown of furazolidone. GF1621.3. Depending on the pathway,
metabolites that still retain the furan
ring with the 5:nitro group may be
formed. Further, metabolites having the
5-nitro group were detected in the urine
of animals treated with furazolidone.
GF-712; GF-751. These metabolites
included the "415" metabolite, of which
the sponsors provide only unsupported
speculation concerning
nonmutagenicity, but which does not
seem to have been investigated. HF-307,
p. 18. Hence, I find that this metabolite
has not been proven safe.
I also find that at least two
metabolites of furazolidone are
mutagenic. The sponsors have cited SF
36 to demonstrate to the contrary.
However, after examining SF36 (pp. 910), I find that two of the acknowledged
metabolites of furazolidonespecifically, aminofuran and
acetamidofuran-are mutagenic. For the
reasons stated at p. 57, infra, I find that
mutagenicity is an indication of
carcinogenicity as well as a separate
health hazard.
The sponsors contend that all the
metabolites in the tissues after the
required 5-day withdrawal period are
harmless because the free metabolites
are water soluble and excreted. Tr. XI72. They claim that the remaining
residues are in the form of adducts,
which are covalently bound forms of
metabolites that are not reactive, and,
therefore, are not of carcinogenic
concern. SX-182; pp. 4, 6-7, 10; SX-180,
pp. 3, 10; SX-181, pp. 4-5, 7-10; HF-307,
pp. 8-10, 12-14. However, my
examination of the evidence contradicts
this position, indicating that not all of
the drug is excreted, and that there are
significant amounts of extractable
residue of furazolidone present in
animal tissue, even-14 days after drug
withdrawal. GF-556; GF-1618.1, p. 11;
GF-1079, pp. 1, 14. This implies that
there are unbound residues in the tissue
or that the bound residues are unstable.
Protein adducts may pose a
toxicological hazard if they are not
stable, according to the evidence. GF1459, pp. 2-3; GF-1545, p. 45. Since the
nature of these residues and their
toxicity were not evaluated, they cannot
be regarded as safe.
The sponsors cite further evidence to
show that, even if the potential adducts
were consumed in treated tissue by
humans, and subsequently hydrolyzed,
no threat would be posed to human

/ Friday, August 23, 1991 / Rules and Regulations

health or safety. HF-307, p. 10. However,
after reviewing the evidence, I find that
hydrolysis in the human digestive
system can free adducts, including
semicarbizide, which has been shown to
be carcinogenic. Tr. XI-30, 92-4.
Residues of furazolidone are clearly
bioavailable. HF-76. Inasmuch as the
identity of all of these residues is not
known, toxicity and carcinogenicity of
these compounds cannot be determined,
and they cannot be considered safe. GF1618.
I also find that not all the-metabolites
of furazolidone are known, and that
their safety, given what we know of the
other metabolites of furazolidone,
cannot be assumed. HF-310, p. 14; GF1617.1, pp. 9, 12; GF-1623.1, p. 22. On the
basis of the factual evidence in the
record, I find that the Winterlin method
of analysis is an unacceptable method
of residue detection until the sponsors
can demonstrate that the marker-the
measured substance-bears a known
relationship to the depletion of the total
drug residue.
Contrary to the sponsors' assertions,
the evidence fails to demonstrate that
furazolidone's metabolites pose no
health risk to the human consumers.
Given all the other evidence in the
record demonstrating that furazolidone
is a carcinogen and that its metabolites
are mutagens, I find that, contrary to the
sponsors' assertions, the metabolites of
furazolidone pose a potential health risk
to human consumers. Because the
sponsors have failed to adduce a
method of detecting furazolidone's total
residues that measures, even indirectly,
the depletion of these residues from
treated animals, I cannot determine that,
under the methods of use specified in
the labeling, no residues of carcinogenic
or toxicological concern remain in the
animal or food products derived from
them.
Accordingly, I find that the NADAs
for furazolidone should be withdrawn
under both the Delaney clause and the
general safety clause, because I have no
reliable method of detecting drug
residues that pose a safety threat to
human consumers.who eat animal
products that may contain furazolidone
residues. Whereas the act requires me to'
consider such residues, it is up to the
sponsors to show that there is a reliable
method to identify and determine the
safety of such residues. They have not
done so.
C. Mutagenicity
I find that furazolidone is a mutagen.
Tr. XII-12-3, 96; SF-36. Mutagenicity is a
scientifically recognized indication of
potential carcinogenicity. HF 104, p. 22. I
agree with Center witness Dr.

Rosenkranz that both furazolidone and
nitrofurazone "have been documented
as mutagenic in systems which are.
highly predictive of cancer-causing
ability." GF-1620.1, p. 013, 1 26. Also, the
genetic damage brought about by a
mutagen is a risk to health by itself,
quite apart from its relation to
carcinogenicity, as-former
Commissioner Jere Goyen found in his
Cyclamate decision (September 16, 1980,
45 FR 61507). Finally, I find that, insofar
as mutagenicity is concerned, the
sponsors have demonstrated no safe
dose of these two nitrofurans. See Tr.
XI-33.
The sponsors claim that, where
furazolidone and/or its metabolites are
shown to be mutagenic, they are only
weakly so and, further, that a weak
mutagen is unlikely to be a carcinogen.
H&C exceptions at 130; SK exceptions at
98. However, I note that nitrofurantoin,
one of the chemicals the sponsors
contended was a weak mutagen but not
a carcinogen, has since been proven to
be an animal carcinogen in a study
submitted for the record by both parties.
See GF-1701. Therefore, based on the
evidence in the record, I find substantial
credible evidence that several of the
known metabolites of furazolidone are
mutagens that must be treated as
carcinogens.
II. Nitrofurazone
A. New Evidence That NitrofurazoneIs
Not Shown To Be Safe
The ALJ found, on the basis of the
evidentiary record before him, that
nitrofurazone is an animal tumorigen,
and, therefore, is not shown to be safe
under the general safety clause. The ALJ
further found that no reliable detection
method has been demonstrated to detect
nitrofurazone-derived residues in edible
animal tissue and that the residues of
nitrofurazone were not shown to be
safe. He concluded that the evidence
before him raised serious scientific
questions about the safety of
nitrofurazone and resulting residues.
I.D., p. 75.
Since the issuance of the I.D., the
record has been reopened to receive a
draft NTP report that finds, on the basis
of state-of-the-art bioassays, that there
is clear evidence that nitrofurazone is an
animal carcinogen. GF-1700. Therefore,
this study both strengthens and
validates the prior evidence of record,
which indicated that nitrofurazone is a
suspect carcinogen.
In the face of overwhelming record
evidence that nitrofurazone is a

Federal Register / Vol. 56, No. 164 / Friday, August 23, 1991 / Rules and Regulations
carcinogen and a tumorigen,7 I find that
new evidence demonstrates that
nitrofurazone is no longer shown to be
safe under the general safety clause.
Thus, the Center has carried its
threshold burden with respect to
nitrofurazone.
B. Residue Detection
The sponsors have offered the same
method of residue detection for
nitrofurazone that they offered for
furazolidone, namely, the Winterlin
method. This method is inadequate to
detect nitrofurazone-derived residues
for the same reason that it is inadequate
to detect furazolidone-derived residues.
The Winterlin method does not detect
residues of any of the metabolites of
nitrofurazone, but only of the parent
drug itself. HF-260; SX-183, pp. 4-5; Tr.
X-11. This omission would not be fatal if
the sponsors had demonstrated that the
depletion of the parent compound from
edible animal tissue bears a known
relationship to the depletion of all
nitrofurazone residues that are
potentially unsafe. However, the
sponsors have produced no such
evidence. In light of this evidentiary
omission, I am unable to determine the
probable consumption of the parent drug
or "of any substance formed in or on
food" (21 U.S.C. 360b(d](2)).as the result
of the use of nitrofurazone in foodproducing animals.
I agree with the Center that no
concentration of the residue of a drug
shown to be a carcinogen, be it in a
parent drug or in its metabolites, can be
shown to be of no carcinogenic concern.
See citations from the Center's main
brief at 82-87; Id. at 76. I find that the
calculation of an acceptable daily intake
(ADI) is inappropriate for a carcinogen.
Tr. XV-15--6. Even if such a calculation
might be appropriate for a carcinogen, I
would have to find that one is not
appropriate for these nitrofurans
because the ADI approach is based
upon observation of a no-observedeffect level, which was not determined
in the Low-Dose Sprague-Dawley rat
study. See citations found in the
Center's main brief at 89.
IV. Other Exceptions
SK excepts to the failure of the ALI to
note that nitrofuraldehyde and 5-nitrofuroics retain the 5-nitro group. SK
Exceptions at 61. I grant this exception
but find that this has no larger
implication with respect to other
conclusions in the I.D. However, I reject
7 There is ample evidence of record that
tumorigens (inducers of benign tumors) can also be
carcinogens (inducers of malignant tumors). GF1700, p. 7; Tr. 111-77-81; Tr. X-112.

SK's contention that these compounds
have low potential for biological activity
because of their low mutagenicity and
rapid oxidation or reduction and
elimination from the animal's body.
First, the relationship between
mutagenicity and carcinogenicity is
qualitative and not quantitative HF-104.
Therefore, low mutagenicity does not
necessarily indicate negligible
carcinogenicity or nonicarcinogenicity.
As to the rapidity of oxidation or
reduction and elimination from the
animal's body, I find that there is of
record no persuasive evidence that
oxidation or reduction rates have any
relationship to the toxicological effects
of the nitrofurans.
2. 1 grant SK's exception (Exceptions
at 62) to the wording of the I.D. at 51,
lines 13-6, concerning whether 4-

ipomeanol or 1-aminopyrine are
metabolites of furazolidone. The
significance of these compounds is that:
(1) They are furans without the 5-nitro
group, and are thus toxic; and (2) aminoaromatic compounds can be activated to
reactive intermediates. Tr. IX-102-3.
Granting this exception does not require
any further amendment to the I.D.
3. As to evidentiary rulings, I affirm
the rulings of the ALJ for the reasons he
stated with one exception. I agree with
SK that the ALJ erroneously struck
portions of the testimony of two
witnesses, Doctors Shriner and Olive, on
grounds that their testimony was
insufficiently supported by citations.
Under the Federal Rules of Evidence, all
relevant evidence is admissible, except
as otherwise provided by law, the
Constitution, or the rules of evidence.
Federal Rules of Evidence, Rule 402.
According to Rule 401, "relevant
evidence," means "evidence having any
tendency to make the existence of any
fact that is of consequence to the
determination of the action more
probable or less probable than it would
be without the evidence." In my view,
the testimony of Doctors Shriner and
Olive, if believed, would have at least
had some tendency to establish SK's
contentions in this proceeding. Further,
under FDA's procedural regulations (21
CFR 12.94) evidence is not made
excludable simply because it contains
either no citations or insufficient
citations. Therefore, I rule that the ALJ
erred in excluding the subject testimony.
The Center's objections should have
been overruled as objections that went
to the weight to be accorded the
testimony, not to its admissibility.
Having overruled the ALJ on this
admissibility question, I nevertheless
find that the testimony of the two
witnesses is entitled to very little weight

41911

as a result of the deficiencies
complained of in the Center's objection.
That is, these witnesses' views are
entitled to little weight because they
were not accompanied by adequate
citations to evidence of record or to any
other supporting literature. For this
reason, although I have considered the
testimony of Doctors Shriner and Olive,
I give it insufficient weight to cause it to
change my mind on any fact in issue in
this proceeding. Though error, the
exclusion was harmless error.
V. Conclusions and Order
The foregoing opinion in its entirety
constitutes my findings of fact and
conclusions of law. Based on the
foregoing discussion, findings, and
conclusions, I affirm the ALJ's initial
decision as corrected and supplemented
by this decision.
Specifically, I conclude that:
(1) New evidence shows that there is
a reasonable basis from which serious
scientific questions may be inferred
about the safety of furazolidone and
nitrofurazone and the residues that
result from their use.
(2) Neither nitrofurazone nor
furazolidone nor their metabolites have
been shown to be safe under the
conditions of use upon the basis of
which the applications were approved
within the meaning of 21 U.S.C.
360b(e)(1)[B).
(3) No reliable detection method has
been demonstrated to be able to detect
nitrofurazone-related residues in edible
tissues when conditions of use approved
in the NADAs are followed.
(4) The residues of nitrofurazone and
furazolidone have not been shown to be
safe.
(5) The Winterlin method of detection
is incapable of measuring the
metabolites of furazolidone. No other
method of detection has been
demonstrated to be able to measure
these metabolites. Hence, no reliable
method of detection has been
demonstrated which is fully adequate to
detect furazolidone-related residues in
edible tissues when conditions of use
approved in the NADAs are followed.
(6) A practical method of detection
capable of detecting both the parent
drug, furazolidone, and its metabolites
does not exist. Therefore, it is
impossible to quantify and qualify the
nature of the residues of furazolidone.
(7) Furazolidone and its metabolites
have been shown by substantial new
evidence to induce cancer in man or
animals as prohibited by 21 U.S.C.
360b(d)(1)(I).
(8) A determination of the
concentration of drug residues

41912.

Federal Register / Vol. 56, No. 164 / Friday, August 23, 1991 / Rules and RLgulations

consisting of the parent drug
furazolidone, and its metabolites that is
of no carcinogenic concern has not been
adequately established.
(9) Under the conditions-of use
specified in the labeling, the actual
concentration of drug residues of
furazolidone, has not been shown to-be.
at or below the level of no carcinogenic.
concern.
Therefore, I order that the approval of
all NADAs for nitrofurazone and
furazolidone listed in this document be
hereby withdrawn pursuant to 21 U.S.C.
360b(d)(i){I} and 360b(e)(1){B). In
addition, I order the removal of 21 CFR
558.262 and 558.370. 1 also order
deletions of all references to
furazolidone and nitrofurazone
contained in 21 CFR 510.515,, 558.4, and
558.15.
List of Subjects
-21 CFR Part510
Administrative practice and
procedure, Animal drugs. Labeling,
Reporting and recordkeeping
requirements.
21 CFR Part558
.

-

Animal drugs, Animal feeds..

-

-

Authority. Secs.512 701 of the Federal
Food. Drug, and Cosmetic Act (21 IJS.C "
360b. 371).

Therefore, under the Federal, Food,
Drug,. and Cosmetic Act, andunder
authority delegated to the Commissioner
of Food and Drugs,. 21 CFR part-510 and
558 -are amended as follows:
-

§ 558A [Amended]
4. Section 558.4 Medicatedfeed
oappicotions-is amended in the Category
II table in paragraph (d) removing the
by
entries for' "Furazolidone" and
"Nitrofurazone."

PART-510-NEW ANIMAL DRUGS
1. The authority citationfor 21 .CFR •
part 510 continues to read as follows:w
-Authority: Secs. 201; 301. 501. 502,, 503, 512,

§ 5568.15 [Amended]
5. Section 558.15 Antibiotic;
nitrofuran, and sulfonamide-drugsin the
feed of animals is amended in the tables
in paragraphs (g)(1) and (g)(2), by
removing the entries for "Hess & Clark
and SmithKline Animal Health
Products."

701, 706 of the Federal Food, Drug, and

Cosmetic Act (21 U.S.C. 321, 331,351. 352, 353,
360b, 371, 378).

§ 510.515 [Amendedl
-2. Section 510.515 Animal feeds
bearingor-containingnew animaldrugs
subject to the provisions of section
512(n) of the act is amended by.
removing paragraphs-(a]{4)-and (a)(5)i;
by removing paragraphs (b)(11). (b)(15).
--(b}(17)iiJ and reserving them; and in the
table in paragraph (c) by removing the
entries for-"8.", "9.",. and "10.," and,
redesignating entries 11 through 14 as 8.
through i1.

-

PART 558-NEW ANIMAL DRUGS FOR.
USE IN ANIMAL FEEDS
3. The authority citation for 21 CFR
-part558 continues to.read as follows:

§ 558.262 [Removed)
6. Section 558.262 Furazolidoneis
removed from subpart B.

-

§ 558.370 [Removed]
.7.
Section 558.370 Nitrofurazoneis
removed from subpart B.,
Dated: August 16,1991.
David A. Kessler,
CommissionerofFoodandDrugs.
[FR Doe. 91-20219 Filed8-22-91, 8:45 am),
CODE 4160-01-M
UMILNG.

-