[Federal Register Volume 63, Number 128 (Monday, July 6, 1998)]
[Rules and Regulations]
[Pages 36344-36362]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 98-17700]


=======================================================================
-----------------------------------------------------------------------

DEPARTMENT OF HEALTH AND HUMAN SERVICES

Food and Drug Administration

21 CFR Part 172

[Docket No. 90F-0220]


Food Additives Permitted for Direct Addition to Food for Human 
Consumption; Acesulfame Potassium

AGENCY: Food and Drug Administration, HHS.

ACTION: Final rule.

-----------------------------------------------------------------------

SUMMARY: The Food and Drug Administration (FDA) is amending the food 
additive regulations to provide for the safe use of acesulfame 
potassium (ACK) as a nonnutritive sweetener in nonalcoholic beverages. 
This action is in response to a petition filed by Hoechst Celanese 
Corp. (Hoechst).

DATES: This regulation is effective July 6, 1998; written objections 
and requests for a hearing by August 5, 1998.

ADDRESSES: Submit written objections to the Dockets Management Branch 
(HFA-305), Food and Drug Administration, 5630 Fishers Lane, rm. 1061, 
Rockville, MD 20852.

FOR FURTHER INFORMATION CONTACT: Patricia A. Hansen, Center for Food 
Safety and Applied Nutrition (HFS-206), Food and Drug Administration, 
200 C St. SW., Washington, DC 20204, 202-418-3093.

SUPPLEMENTARY INFORMATION: 

Table of Contents

I. Introduction
II. Evaluation of Safety
III. Evaluation of the Safety of the Petitioned Use of the Additive
    A. ACK--Background
    B. ACK--New Information
    C. Methylene Chloride
    D. Special Conditions Relevant to Use in Nonalcoholic Beverages
      1. Hydrolysis Products--Consumer Exposure
      2. Hydrolysis Products--Evaluation of Toxicological 
Information
        a. Acetoacetamide-N-sulfonic acid (AAS)
          i. Genetic toxicity testing
          ii. Subchronic toxicity studies in rats and monkeys
          iii. Reproduction and developmental toxicity study in rats
          iv. Assessment of AAS
        b. Acetoacetamide (AAA)
          i. Genetic toxicity testing
          ii. Short-term and subchronic toxicity studies in rats, 
rabbits, and dogs
          iii. Developmental toxicity study in rabbits
          iv. Assessment of AAA--nonthyroid endpoints
          v. Assessment of AAA--thyroid endpoints
          vi. Consideration of whether more testing of AAA is 
necessary
    E. Summary of FDA's Safety Evaluation
IV. Response to Comments
    A. Summary of Comments
      1. Center for Science in the Public Interest's (CSPI's) First 
Submission
      2. CSPI's Second Submission
      3. CSPI's Third Submission
      4. CSPI's Fourth Submission
      5. Hoechst's Submission
      6. Other Submissions
    B. Analysis of Specific Issues Raised in the Comments
      1. AAA Test Results
      2. ACK Test Results
        a. The second rat study
          i. Issues raised previously
          ii. Issues not raised previously
        b. The mouse study
          i. Issues raised previously
          ii. Issues not raised previously
        c. The first rat study
           Issues raised previously
    C. Summary of FDA's Response to Comments
V. Conclusion of Safety
VI. Environmental Impact
VII. Paperwork Reduction Act
VIII. Objections
IX. References

I. Introduction

    In a notice published in the Federal Register of July 30, 1990 (55 
FR 30983), FDA announced that a food additive petition (FAP 0A4212) had 
been filed by Hoechst Celanese Corp. (Hoechst), Route 202-206 North, 
Somerville, NJ 08876, proposing that Sec. 172.800 Acesulfame potassium 
(21 CFR 172.800) be amended to provide for the safe use of acesulfame 
potassium (ACK) as a nonnutritive sweetener in nonalcoholic beverages, 
including beverage bases. (Recently, Hoechst has reorganized; the 
division of Hoechst now responsible for ACK is known as Nutrinova, 
Inc., 25 Worlds Fair Dr., Somerset, NJ 08873.) The present petition 
contains data and other information relevant to the safety of ACK under 
the proposed conditions of use; the present petition also relies on 
certain data and information contained in previous petitions for ACK.
    FDA's food additive regulations were first amended to permit the 
use of ACK on July 28, 1988 (53 FR 28379, the ``dry uses final rule''), 
in response to a petition filed by Hoechst. In its original evaluation 
of the safety of ACK, FDA concluded that a review of animal feeding 
studies showed that there is no association between neoplastic disease 
(cancer) and consumption of this additive (53 FR 28379 at 28380 and 
28381). The agency further concluded that ACK was safe under the 
conditions of use proposed in the initial petition, and amended its 
food additive regulations to permit the use of the sweetener.
     Following publication of the dry uses final rule, the agency 
received timely objections from the Center for Science in the Public 
Interest (CSPI). CSPI submitted four separate objections, two of which 
asserted that the long-term studies of ACK in rodents were inadequate 
to evaluate ACK's potential carcinogenicity, and two of which asserted 
that certain of these studies showed that the additive was potentially 
carcinogenic. CSPI requested a stay of the regulation and also 
requested a hearing on each of its objections. FDA, after careful 
consideration of CSPI's objections, found that none of the objections 
raised issues of fact that justified granting a

[[Page 36345]]

hearing or otherwise provided a basis for revoking the regulation. Thus 
FDA denied both the request for a stay of the regulation and a hearing, 
and confirmed the effective date of the regulation. The agency 
published a detailed response to CSPI's objections in the Federal 
Register of February 27, 1992 (57 FR 6667).
    Since its initial approval decision on the use of ACK, FDA has 
approved the following additional uses for ACK in response to 
petitions: In baked goods and baking mixes, including frostings, 
icings, and fillings for baked goods; in yogurt and yogurt-type 
products; in frozen and refrigerated desserts; in sweet sauces, 
toppings, and syrups; and in alcoholic beverages (59 FR 61538, 59 FR 
61540, and 59 FR 61543, December 1, 1994, and 60 FR 21700, May 3, 
1995). No objections were received in response to the December 1, 1994, 
final rule. However, CSPI filed timely objections to the agency's May 
3, 1995, final rule authorizing the use of ACK in alcoholic beverages 
(60 FR 21700). The agency's response to those objections is published 
elsewhere in this issue of the Federal Register.
    With respect to the present petition, Hoechst's original submission 
contained data and information from several toxicity studies of ACK, as 
well as data and information regarding the stability of ACK in aqueous 
solutions.\1\ Because hydrolysis of ACK can occur under certain 
conditions, the petitioner also conducted toxicity studies of the 
principal hydrolysis products of ACK.
---------------------------------------------------------------------------

    \1\ Stability studies of ACK in aqueous solutions were also 
submitted in the original petition for ACK.
---------------------------------------------------------------------------

    In response to an issue raised by FDA's review, Hoechst submitted 
additional information regarding ACK hydrolysis products, including a 
report prepared by a panel of experts in various scientific disciplines 
who independently evaluated the results of certain toxicity studies of 
the ACK hydrolysis products. Hoechst also submitted an indepth analysis 
of the potential health risk from one of the ACK hydrolysis products, 
acetoacetamide (AAA). FDA's Center for Food Safety and Applied 
Nutrition (CFSAN) conducted its own indepth analysis of the data and 
information on AAA, and, in reaching a final decision on this issue, 
also obtained the advice of additional experts from within and from 
outside the agency.
    FDA notes that CSPI has submitted comments on the present petition 
for use of ACK in nonalcoholic beverages, and has transmitted comments 
on that petition from other interested parties as well. Further, 
Hoechst has transmitted additional comments from two of these same 
parties. Several other comments were also received. The agency's 
response to all comments on the present petition is presented in 
section IV of this document.

II. Evaluation of Safety

    Under the general safety standard of the Federal Food, Drug, and 
Cosmetic Act (the act) (21 U.S.C. 348(c)(3)(A)), a food additive cannot 
be approved for a particular use unless a fair evaluation of the data 
available to FDA establishes that the additive is safe for that use. 
FDA's food additive regulations in Sec. 170.3(i) (21 CFR 170.3(i)) 
define safe as ``a reasonable certainty in the minds of competent 
scientists that the substance is not harmful under the intended 
conditions of use.''
    The food additives anticancer, or Delaney, clause of the act (21 
U.S.C. 348(c)(3)(A)) provides that no food additive shall be deemed 
safe if it is found to induce cancer when ingested by man or animal. 
Importantly, however, the Delaney clause applies to the additive itself 
and not to impurities in the additive. That is, where an additive 
itself has not been shown to cause cancer, but contains a carcinogenic 
impurity, the additive is properly evaluated under the general safety 
standard using risk assessment procedures to determine whether there is 
a reasonable certainty that no harm will result from the intended use 
of the additive (Scott v. FDA, 728 F.2d 322 (6th Cir. 1984)).

III. Evaluation of the Safety of the Petitioned Use of the Additive

A. ACK--Background

    As previously noted, FDA's original evaluation of the safety of ACK 
established that there was no association between neoplastic disease 
(cancer) and consumption of this additive (53 FR 28379 at 28380 and 
28381). That evaluation also established a lifetime-averaged acceptable 
daily intake (ADI) for ACK of 15 milligrams per kilogram of body weight 
per day (mg/kg bw/d), equivalent to 900 mg per person per day (mg/p/d).

B. ACK--New Information

    In the present petition, Hoechst included several ACK toxicity 
studies that had been conducted since the agency's original evaluation 
of the safety of this additive. These included studies on mutagenicity, 
antigenicity, and potential for dermal and eye irritation; an acute 
toxicity study in fish; and a subchronic toxicity study in diabetic 
rats.
    The mutagenicity studies demonstrated that ACK is not mutagenic at 
histidine loci in Salmonella typhimurium or at a tryptophan locus in 
Escherichia coli. These results are consistent with the negative 
results of the mutagenicity and genetic toxicity studies previously 
considered by FDA in its original evaluation of the safety of ACK. The 
results of all the ACK genetic toxicity tests establish that ACK is not 
genotoxic.
    The results of the other ACK toxicity studies listed above did not 
show toxicologically significant ACK-related adverse effects. 
Importantly, these ACK toxicity studies contain no new information that 
would change the agency's previous conclusion that there is no 
association between neoplastic disease and consumption of this 
additive. Thus, FDA has evaluated the safety of the petitioned use of 
ACK in nonalcoholic beverages under the general safety standard, 
considering all available data.
    In determining whether the proposed use of an additive is safe, FDA 
considers, among other things, whether an individual's lifetime-
averaged estimated daily intake (EDI) of the additive will be less than 
the ADI established from toxicological information. Importantly, the 
new studies on ACK listed above do not contain any new information that 
would cause the agency to alter the previously determined ADI for ACK. 
Thus, FDA concludes that the ADI for ACK is 15 mg/kg bw/d (equivalent 
to 900 mg/p/d). The present petition contains information regarding 
dietary consumption of ACK-containing food products, including 
nonalcoholic beverages, and the agency has considered consumer exposure 
to ACK resulting from its use in nonalcoholic beverages, as well as all 
currently listed uses. FDA has calculated the mean EDI from these 
combined uses to be 1.6 mg/kg bw/d, which is equivalent to 96 mg/p/d; 
and the 90th percentile EDI from these combined uses to be 3.0 mg/kg 
bw/d, which is equivalent to 180 mg/p/d (Ref. 1). These levels of 
dietary exposure to ACK, which represent measures of the average and 
the high chronic intake, respectively, are both well below the ADI.

C. Methylene Chloride

    Residual amounts of reactants and manufacturing aids are commonly 
found as contaminants in chemical products, including food additives. 
In its evaluation of the safety of ACK, FDA reviewed both the safety of 
the additive

[[Page 36346]]

and the safety of the chemical impurities that may be present in the 
additive from the manufacturing process.
    In the current manufacturing process for ACK, methylene chloride, a 
carcinogenic chemical, is used as a solvent in the initial 
manufacturing step. Subsequently, the product is neutralized, stripped 
of methylene chloride, and recrystallized from water. Data submitted by 
the petitioner show that methylene chloride could not be detected in 
the final product at a limit of detection of 40 parts per billion 
(ppb).
    FDA has previously discussed the significance of the use of 
methylene chloride in the production of ACK. The agency incorporates 
those discussions, published in the Federal Register of December 1, 
1994 (59 FR 61538, 59 FR 61540, and 59 FR 61543) and of May 3, 1995 (60 
FR 21700), in full, into the agency's safety determination on the 
present petition.
    Specifically, in evaluating the safety of the uses of the additive 
that are currently listed, FDA concluded, using risk assessment 
procedures, that the estimated upper-bound limit of individual lifetime 
risk from the potential exposure to methylene chloride resulting from 
these uses of ACK, together with the petitioned use of ACK in 
nonalcoholic beverages, is 2.6 x 10-11, or less than 3 in 
100 billion. The agency also concluded that, because of the numerous 
conservative assumptions used in calculating this estimated upper-bound 
limit of risk, this upper-bound limit would be expected to be 
substantially higher than any actual risk (59 FR 61538 at 61539, 59 FR 
61540 at 61542, 59 FR 61543 at 61544, and 60 FR 21700). FDA has 
received no new information that would change the agency's previous 
conclusion. Therefore, the agency concludes that there is a reasonable 
certainty of no harm from the exposure to methylene chloride that might 
result from the proposed use of ACK in nonalcoholic beverages.
    In conducting its evaluation, the agency also considered whether a 
specification is necessary to control the amount of potential methylene 
chloride impurity in ACK. At that time, FDA concluded that there is no 
reasonable possibility that methylene chloride will be present in 
amounts that present a health concern, and that there would thus be no 
justification for requiring manufacturers to monitor compliance with a 
specification (59 FR 61538 at 61539, 59 FR 61540 at 61542, 59 FR 61543 
at 61544, and 60 FR 21700). Because no new information has been 
received that would change FDA's previous conclusion regarding the need 
for a specification, the agency affirms its prior determination that a 
specification for methylene chloride impurity in ACK is unnecessary.

D. Special Conditions Relevant to Use in Nonalcoholic Beverages

    The use of ACK as a nonnutritive sweetener in nonalcoholic 
beverages may subject the sweetener to conditions other than those 
considered in the evaluation of the currently listed uses of this 
additive. FDA has evaluated data in the present petition and other 
information regarding the stability of ACK under a variety of 
conditions that characterize the proposed use in nonalcoholic 
beverages. Based on these data and information, the agency concludes 
that ACK is stable under almost all circumstances expected to be 
encountered for the proposed use in nonalcoholic beverages.
    However, FDA has determined that there is a limited possibility 
that some nonalcoholic beverages could be stored under conditions that 
could lead to the formation of ACK hydrolysis products. Specifically, 
small amounts of hydrolysis products may be formed in highly acidic 
aqueous food products (which would include some, though not all, 
nonalcoholic beverages) under conditions of prolonged storage at 
elevated temperatures. As part of its safety evaluation, FDA has 
reviewed toxicological data and supporting information regarding the 
hydrolysis products of ACK, as well as estimates of human dietary 
exposure to the hydrolysis products. The substantive aspects of the 
agency's safety assessment of the hydrolysis products, as they relate 
to the use of ACK in nonalcoholic beverages, are discussed in detail in 
sections III.D.1 and 2 of this document.
1. Hydrolysis Products--Consumer Exposure
    Both the present petition and the petition supporting the initial 
approval of ACK contain studies of the stability of ACK in aqueous 
solutions. These studies show that ACK hydrolyzes, in strongly acidic 
or strongly basic aqueous solutions, to acetoacetamide-N-sulfonic acid 
(AAS). AAS subsequently hydrolyzes to acetoacetamide (AAA). The AAA 
that is formed is also subject to hydrolysis; the eventual endproducts 
are acetone, carbon dioxide, and ammonia. Data and other information 
submitted by the petitioner and evaluated by the agency establish that 
both AAS and AAA are transient intermediates in the overall ACK 
hydrolysis pathway and that no significant buildup of AAS or AAA will 
occur in ACK-sweetened nonalcoholic beverages.
    Studies in the two petitions also establish that hydrolysis of ACK 
is dependent on two other factors in addition to pH: Time and 
temperature. Prolonged storage at elevated temperatures is required to 
produce detectable amounts of AAS and, particularly, its byproduct, 
AAA, even in test solutions containing over 100 times the amount of ACK 
that would ordinarily be used in a nonalcoholic beverage. Specifically, 
data in the petition show that such a concentrated, buffered, 
carbonated solution of pH 3.0 (representative of the lower end of the 
pH range for carbonated diet soft drinks), after storage at 20  deg.C 
(68  deg.F) for 8 weeks, contained AAS at a level of 0.35 percent of 
the original ACK level. Even with a sensitive analytical method (limit 
of detection, circa (ca.) 1 ppb, corresponding to 0.001 percent of the 
original ACK level), no AAA was detected in this system. More severe 
storage conditions were required to produce detectable levels of AAA 
(e.g., 8 weeks storage at 30  deg.C (86  deg.F) or 50 weeks storage at 
20  deg.C).
    The combination of conditions necessary to produce measurable 
amounts of hydrolysis products in beverages (i.e., low beverage pH and 
extended storage at high temperatures) is not expected to be frequently 
encountered. The stability studies also establish that AAA and AAS will 
not build up in beverages over time. Accordingly, FDA believes that any 
consumer exposure to AAA and AAS from consumption of ACK-sweetened 
nonalcoholic beverages will be at extremely low levels and also both 
intermittent and infrequent.
    Nevertheless, using data from the stability studies and other 
information regarding consumption patterns, FDA has estimated a 
potential lifetime-averaged ``daily'' dietary intake of ACK hydrolysis 
products that might result from consumption of ACK-sweetened 
nonalcoholic beverages. In its calculations, the agency has 
deliberately incorporated several assumptions that, taken together, 
will produce an estimated ``daily'' intake that is likely to be an 
overestimate rather than an underestimate. First, FDA has assumed that 
all nonalcoholic beverages ingested by consumers will have been 
sweetened only with ACK, that ACK will be used at the highest levels 
characteristic of each type of nonalcoholic beverage, and that the 
consumer will have ingested such beverages at the 90th percentile 
consumption level. Second, FDA has assumed certain values for beverage 
pH,

[[Page 36347]]

storage time, and storage temperature that are also likely to produce 
an overestimate of the ``daily'' intake of ACK hydrolysis products. The 
basis for the agency's particular choice of beverage pH, storage time, 
and storage temperature is discussed in more detail in the next two 
paragraphs.
    FDA has chosen to use a pH of 3.0 in its analysis because this pH 
is representative of the lower end of the range in which beverages 
containing nonnutritive sweeteners are formulated. The agency has 
chosen to use a storage time of 8 weeks because FDA considers 8 weeks 
to be representative of a storage period that is significantly longer 
than the average storage period for nonalcoholic beverages. Data in the 
petition and in the agency's files show that ca. 90 percent of diet 
cola (representative of beverages formulated at low pH) is sold within 
8 weeks of bottling; these data also show that even when additional 
flavor categories are considered, ca. 90 percent of nonalcoholic 
beverages are still sold within 9.5 weeks of bottling, with an average 
time from bottling to sale of just under 4 weeks (Ref. 2).
    With respect to temperature, FDA has chosen to use 20  deg.C in its 
analysis because this temperature is representative of the high end of 
the range of in-home or in-store storage temperatures, when periods of 
both refrigerated and room temperature storage are taken into 
account.\2\ The agency also reviewed climate data for different 
geographical locations in the United States, which were chosen to cover 
the range of possible temperature extremes for beverages stored under 
ambient conditions (no temperature control). This review shows that few 
locations have annual average temperatures above 20  deg.C (Ref. 2). 
Accordingly, for all of the foregoing reasons, the agency has used 20 
deg.C as representative of the temperature conditions likely to be 
encountered over an extended storage period.
---------------------------------------------------------------------------

    \2\ FDA also considered the effect of extreme temperature 
conditions on dietary exposure to ACK hydrolysis products (see Ref. 
2). However, the agency has concluded that, for several reasons, it 
is highly unlikely that beverages stored under extremely high 
temperatures for extended periods of time would be consumed on a 
continued basis. First, most in-home or retail storage is under 
refrigeration or other climate-controlled conditions. Second, it is 
a common and usual practice in the industry to discard diet 
beverages that have been stored under extreme conditions (e.g., 50 
to 55  deg.C, equivalent to 120 to 130  deg.F) because the 
artificial sweeteners currently in use undergo significant 
decomposition that results in an unpalatable product. FDA expects 
that this practice would also be applied to beverages sweetened with 
ACK because the decomposition of ACK that occurs under such extreme 
conditions also results in an unpalatable product. Finally, 
consumers do not customarily store nonalcoholic beverages under 
extreme conditions for lengthy periods, and would not be expected to 
habitually consume the unpalatable products that result from 
extended storage at extremely high temperatures.
---------------------------------------------------------------------------

    FDA has calculated estimated dietary exposure to AAS and AAA based 
upon data reflecting the foregoing assumptions regarding beverage 
formulation and storage conditions (see Ref. 2). The agency concludes 
that, for the 90th percentile consumer of ACK-sweetened nonalcoholic 
beverages, exposure to AAS would be no more than 2.5 micrograms 
(g)/kg bw/d, which is equivalent to 0.15 mg/p/d. In estimating 
consumer exposure to AAA, the agency incorporated an additional 
conservative assumption: that AAA would be present at a level 
corresponding to one-half the limit of detection (Ref. 3), even though 
it was not actually detected. The agency concludes that, for the 90th 
percentile consumer of ACK-sweetened nonalcoholic beverages, exposure 
to AAA would be no more than 3.3 nanograms (ng)/kg bw/d, which is 
equivalent to 0.2 g/p/d.
2. Hydrolysis Products--Evaluation of Toxicological Information
    In support of the safety of ACK for use as a nonnutritive sweetener 
in nonalcoholic beverages, the petitioner submitted toxicity studies of 
AAS and AAA, the two principal hydrolysis products of ACK. The agency's 
evaluation of these toxicological data and other related information 
follows.
    a. Acetoacetamide-N-sulfonic acid (AAS). Hoechst submitted a set of 
toxicity studies of AAS in support of the safety of the proposed use of 
ACK in nonalcoholic beverages including: Short-term tests for genetic 
toxicity; acute, short-term and subchronic studies in rats; a 
subchronic study in dogs; short-term and subchronic studies in monkeys; 
an acute study in humans; a reproduction and developmental toxicity 
study in rats; and metabolism studies in rats and humans. The key 
studies of AAS relevant to FDA's safety decision regarding the 
petitioned use of ACK are discussed in the next sections of this 
document.
    i. Genetic toxicity testing. AAS was tested in several in vitro and 
in vivo genetic toxicity tests. In the absence of bioassay data, such 
tests are often used to predict the carcinogenic potential of the test 
compound.
    AAS was not mutagenic at histidine loci in Salmonella typhimurium 
(Ames test), at a tryptophan locus in Escherichia coli, nor at the 
HGPRT locus in V79 cells treated in vitro. AAS did not induce 
unscheduled deoxyribonucleic acid (DNA) synthesis in strain A 549 human 
cells exposed in vitro. Finally, AAS was not clastogenic in V79 cells 
exposed in vitro nor in bone marrow cells of NMRI mice. The agency 
concludes that results of these tests establish that AAS is not 
genotoxic.
    ii. Subchronic toxicity studies in rats and monkeys. The petitioner 
submitted the results of a subchronic toxicity study in which AAS was 
administered in the diet to 30 Wistar rats/sex/group at dose levels 
equivalent to 0, 800, 2,000, or 5,000 mg/kg bw/d for 90 days. Twenty 
rats/sex/group were sacrificed at the end of the dosing period. The 
remaining ten rats/sex/group were designated as ``recovery'' animals; 
that is, there was an interval of approximately 1 month between the 
time dosing ended and the time of sacrifice for these animals.
    Increased relative kidney weights and decreased relative pituitary 
weights were observed in high-dose female rats. The mid- and high-dose 
groups (2,000 and 5,000 mg/kg bw/d, respectively) of male and female 
rats had softer feces, decreased body weight gain, and dose-related 
increases in feed consumption compared to controls. Other AAS-related 
effects observed in the animals in the mid- and high-dose groups 
included increased urine pH, and changes in various clinical chemistry 
parameters, some of which changes resolved by the end of the recovery 
period. Certain changes in the caecum were also observed; however, 
these effects had also resolved by the end of the recovery period, and 
were judged by FDA to be a probable physiological adaptation to osmotic 
changes in the gastrointestinal tract. Based on these data, FDA 
concludes that the no-observed-effect level (NOEL) from this study is 
800 mg AAS/kg bw/d, the lowest dose level tested in this study (Ref. 
4).
    The petitioner also submitted the results of a subchronic toxicity 
study of AAS in Cynomologous monkeys. In this study, four monkeys/sex/
group were administered gavage doses of 0, 100, 315, or 1,000 mg AAS/kg 
bw/d for 13 weeks. Marginal decreases in the absolute and relative 
weights of various organs in animals of the mid- and high-dose groups 
were observed; however, FDA does not consider these effects to be of 
toxicological significance because of the lack of corroborative 
evidence of organ toxicity. The only toxicologically significant effect 
observed in this study was a dose-related increase in incidence and 
severity of diarrhea in the mid- and high-dose groups. Thus, FDA 
concludes that the NOEL for AAS from this study

[[Page 36348]]

is 100 mg/kg bw/d, the lowest dose level tested (Ref. 4).
    iii. Reproduction and developmental toxicity study in rats. The 
petitioner submitted the results of a two-generation reproduction study 
with a teratology phase conducted in Sprague-Dawley rats. In this 
study, AAS was administered in the diet to 25 rats/sex/group of the P- 
and F1-generation at dose levels equivalent to 0, 164, 492, or 1,780 mg 
AAS/kg bw/d. No adverse effects on reproduction or developmental 
parameters were observed at any dose level in this study. Thus, FDA 
concludes that the NOEL for this study is 1,780 mg AAS/kg bw/d, the 
highest dose used in the study (Ref. 4).
    iv. Assessment of AAS. No adverse AAS-related effects were observed 
at 800 mg/kg bw/d in the subchronic rat study, at 100 mg/kg bw/d in the 
subchronic monkey study, and at 1,780 mg/kg bw/d and lower in the 
reproduction/teratology study in rats. The agency has no safety 
concerns about AAS at its anticipated level of intake (less than 2.5 
g/kg bw/day) because of the substantial margin between this 
level and the levels at which no adverse effects were observed in these 
studies (a margin of at least 40,000).
    b. Acetoacetamide (AAA). Hoechst submitted a set of toxicity 
studies of AAA in support of the safety of ACK for use in nonalcoholic 
beverages, including short-term tests for genetic toxicity; an acute 
study, two short-term studies, and a subchronic study in rats; an acute 
and two short-term studies in dogs; a subchronic study in rabbits; 
metabolism studies in rats, dogs, hamsters, and humans; a developmental 
toxicity study in rabbits; and several other studies. The key studies 
of AAA relevant to FDA's safety decision regarding the petitioned uses 
of ACK are discussed in detail below.
    i. Genetic toxicity testing. AAA was tested in several in vitro and 
in vivo genetic toxicity tests. As noted, in the absence of bioassay 
data, such tests are often used to predict the carcinogenic potential 
of the test compound.
    AAA was not mutagenic at the HGPRT locus in V79 cells treated in 
vitro nor at histidine loci in Salmonella typhimurium (Ames test). AAA 
was not clastogenic in V79 cells exposed in vitro nor in bone marrow 
cells of NMRI mice. In addition, AAA did not induce unscheduled DNA 
synthesis in strain A 549 human cells exposed in vitro. The agency 
concludes that the results of these tests establish that AAA is not 
genotoxic.\3\
---------------------------------------------------------------------------

    \3\ The petitioner also submitted results of genetic toxicity 
tests of -hydroxybutyramide (BHB), the principal metabolite 
of AAA in humans. The Ames test of BHB was well conducted and showed 
that BHB is not mutagenic. Although several of the other genetic 
toxicity tests of BHB had deficiencies, none of these tests 
indicated that BHB is genotoxic.
---------------------------------------------------------------------------

    ii. Short-term and subchronic toxicity studies in rats, rabbits, 
and dogs. The petitioner submitted the results of one subchronic (90-
day) and two short-term toxicity studies of AAA in rats. One short-term 
(30-day) study was designed to determine appropriate doses for the 
subsequent subchronic study. The second short-term (14-day) study was 
designed as a preliminary mechanistic study; the second short-term 
study is discussed in detail in section III.D.2.b.v of this document.
    In the subchronic study, AAA was administered in the diet to 15 SPF 
Wistar rats/sex/group at dose levels equivalent to 0, 24, 157, 794, or 
4,300 mg/kg bw/d for 13 weeks. The following AAA-related adverse 
effects were identified in the subchronic rat study: (1) Reduced body 
weights of males and females in the highest dose group over the entire 
study; (2) anemia in female rats in the highest dose group and male 
rats in the two highest dose groups; (3) increased numbers of both 
males and females with centrilobular fatty liver in the highest dose 
group; (4) increased group mean relative liver weights for male and 
female rats in the highest dose group; as well as (5) various adverse 
effects on the thyroid, which are discribed in the next paragraph.
    The adverse effects on the thyroid observed in the subchronic rat 
study of AAA were: (1) Dose-related increases in the numbers of males 
and females with grossly enlarged thyroids; (2) increased relative 
thyroid weights for mid- and high-dose males and females; (3) dose-
related increases in the numbers of males and females with follicular 
cell hypertrophy and hyperplasia; and (4) thyroid adenomas in one male 
rat in each of the two highest dose groups. No hypertrophy or 
hyperplasia was associated with enlarged thyroids in controls or in 
animals in the lowest dose group (24 mg/kg bw/d).
    With respect to endpoints in organs other than the thyroid, no 
adverse toxicological effects were observed at doses corresponding to 
157 mg/kg bw/day and lower. However, based on the gross and 
histopathological findings in the thyroid, FDA concludes that the NOEL 
from the subchronic rat study is 24 mg AAA/kg bw/d, the lowest dose 
tested in this study.
    The petitioner also submitted the results of a subchronic study of 
AAA in albino Himalayan rabbits. In this study, six rabbits/sex/group 
were administered 0, 1,200, 6,000, or 30,000 mg AAA/kg drinking water/
day (equivalent to 0, 96, 499, or 2,192 mg AAA/kg bw/d for male 
rabbits, and to 0, 93, 560, or 2,763 mg AAA/kg bw/d for female 
rabbits). The following effects were observed: (1) Significantly 
increased testes weights and signs of focal tubular hypospermatogenesis 
in the testes of all high-dose males; (2) significantly increased 
thyroid weights in high-dose males and females; and (3) thyroid 
follicular cell hypertrophy and hyperplasia in all high-dose males and 
females. One mid-dose female and one high-dose female in this study had 
grossly enlarged thyroids; the mid-dose female also had a thyroid 
follicular cyst that may have been part of a hyperplastic response.
    With respect to endpoints in organs other than the thyroid, no 
adverse toxicological effects were observed at doses corresponding to 
499 mg/kg bw/day and lower. However, based on the evidence that the 
thyroid is a target organ for AAA-related toxicity and the finding of 
possible thyroid hyperplasia in one female in the mid-dose group, FDA 
concludes that the NOEL for AAA in rabbits is 93 mg/kg bw/d, the lowest 
dose tested in females in this study (Ref. 4).
    The petitioner submitted the results of two short-term (14-day) 
studies of AAA in dogs. In the first short-term study, two dogs/sex/
group were gavaged with 0, 100, 500, or 2,500 mg AAA/kg bw/d for 14 
days. Thyroid follicular cell hyperplasia was observed in males and 
females in all dose groups.
    Because adverse effects were observed at all dose levels in the 
first study, the petitioner performed a second short-term (14-day) dog 
study using lower doses. In the second study, three dogs/sex/group were 
gavaged with 0, 4, 20, or 100 mg AAA/kg bw/d for 14 days; at the end of 
the dosing period two males and females from each group were 
sacrificed. The remaining male and female in each group were designated 
as ``recovery'' animals; that is, there was an interval of 
approximately 1 month between the time dosing ended and the time of 
sacrifice for these two animals. In this study, two of the males in the 
high-dose group developed thyroid follicular hyperplasia; no other 
males and no females in this study were reported to have thyroid 
abnormalities. However, of the two high-dose males that developed 
thyroid follicular hyperplasia, one was a ``recovery'' animal, 
indicating that the effect of AAA on the thyroid had persisted for 1 
month after dosing ended. In an effort to identify a possible mechanism 
for AAA's action on the thyroid in the second dog study, the 
investigators

[[Page 36349]]

measured serum levels of thyroid hormones T3 and T4 at the end of the 
study; no compound-related changes in serum T3 or T4 levels were 
observed. (The investigators did not measure levels of thyroid 
stimulating hormone (TSH).)
    FDA concludes that the results of the short-term and subchronic 
toxicity studies in rats, rabbits, and dogs demonstrate that AAA has a 
proliferative effect on the thyroid (i.e., diffuse follicular cell 
hypertrophy and hyperplasia). The agency's assessment of the 
significance of the observed thyroid lesions is discussed in detail in 
section III.D.2.b.v of this document.
    iii. Developmental toxicity study in rabbits. The petitioner 
submitted an embryotoxicity study of AAA in Chinchilla rabbits in which 
groups of 16 rabbits were gavaged with 0, 100, 300, or 1,000 mg AAA/kg 
bw/d on days 6 through 18 of pregnancy. FDA has determined that there 
were no toxicologically significant effects of AAA on reproductive or 
developmental parameters in this study; thus, the NOEL for reproductive 
and developmental effects is 1,000 mg AAA/kg bw/d, the highest dose 
used in this study (Ref. 4).
    iv. Assessment of AAA--nonthyroid endpoints. For organs other than 
the thyroid, no AAA-related adverse effects were observed at 157 mg/kg 
bw/d and lower in the subchronic rat study, at 499 mg/kg bw/d and lower 
in the subchronic rabbit study, and at 1,000 mg/kg bw/d and lower in 
the developmental toxicity study in rabbits. With respect to endpoints 
in organs other than the thyroid, the agency has no safety concerns 
about AAA at its anticipated level of intake (less than 3.3 ng/kg bw/
day) because of the substantial margin between this level and the 
levels at which no adverse effects were observed in the studies 
discussed previously (a margin of at least 5 million).
    v. Assessment of AAA--thyroid endpoints. No adverse AAA-related 
effects on the thyroid were observed at 24 mg/kg bw/day in the 
subchronic rat study, at 93 mg/kg bw/day in the subchronic rabbit 
study, and at 20 mg/kg bw/day and lower in the second short-term dog 
study. Although the study results permit FDA to identify NOEL's for 
certain thyroid endpoints in the rat and rabbit subchronic studies,\4\ 
the major histological findings in these studies, thyroid follicular 
cell hypertrophy and hyperplasia, raise a question regarding the 
possible tumorigenic activity of AAA. Thyroid follicular cell 
hypertrophy and hyperplasia were also observed at similar levels of AAA 
administration in the dog studies, which studies were of even shorter 
duration. The pronounced thyroid follicular cell hypertrophy and 
hyperplasia observed in rats, rabbits, and dogs, considered together 
with the occurrence of thyroid adenomas in two males in the subchronic 
rat study, suggest that AAA might induce thyroid tumors if administered 
in long-term oral studies (see Refs. 2 and 4).
---------------------------------------------------------------------------

    \4\ In reaching a safety decision on a food additive, FDA 
typically uses NOEL's determined from studies of at least 90 days 
duration (a subchronic study) and uses the term ``NOEL'' to refer 
specifically to the no-observed-effect levels determined from such 
studies. Results from studies in which animals are exposed for 
shorter test periods are typically used for different purposes 
(e.g., to gather information for use in designing longer studies). 
The short-term studies in dogs and rats (14 days) are too short to 
determine a subchronic NOEL.
---------------------------------------------------------------------------

    In response to FDA's concerns regarding AAA's thyroid effects, the 
petitioner initially argued that application of an appropriate safety 
factor to the lowest NOEL for thyroid endpoints was a suitable 
approach, despite the possible tumorigenic activity of AAA. Hoechst 
maintained that the dose-related hypertrophy and hyperplasia of the 
thyroid follicular cells and, in a 90-day study, the progression of 
some cells to adenomas was consistent with a typical pattern of 
morphological changes clearly associated with sustained, elevated 
levels of TSH,\5\ particularly in the rat. Hoechst also maintained that 
AAA was most likely to act on the thyroid gland by inhibiting the 
enzyme thyroperoxidase in follicular cells. Thyroperoxidase is required 
for synthesis of T3 and T4 in the thyroid; therefore, inhibiting this 
enzyme would lead to a reduction in the levels of T3 and T4 and, 
consequently, increased serum levels of TSH (see Refs. 5 and 6). As 
support for this hypothesis, Hoechst referenced an extensive body of 
scientific literature linking thyroperoxidase inhibition (and 
consequent elevated TSH levels) by other compounds to thyroid lesions 
that are similar in type, severity, and timecourse of development, to 
the thyroid lesions observed in the short-term and subchronic studies 
of AAA summarized previously in this document. Hoechst asserted that 
progression of the hypertrophy and the hyperplasia associated with AAA 
would be dependent on continued or chronic stimulation of the thyroid 
gland by TSH, again drawing upon comparisons with other compounds whose 
similar effects on the thyroid were mediated by chronic TSH 
stimulation.\6\
---------------------------------------------------------------------------

    \5\ Iodine is taken up by the thyroid and converted to the 
thyroid hormone thyroxine, also known as T4 (which contains four 
iodine atoms) or to tri-iodothyronine, otherwise known as T3 (which 
contains three iodine atoms). Thyroid hormone production and release 
into circulation are stimulated by TSH released by the pituitary in 
response to decreases in circulating levels of T3 and T4. The 
biological functions of T4 and T3 are similar. The thyroid hormones 
are primarily metabolized in the liver and, to a lesser extent, in 
the kidneys. T4 can be converted to T3 (biologically active) or to 
reverse T3 (inactive), and then to di-iodothyronine (DIT).
    Thyroid hypertrophy, hyperplasia and neoplasia can be caused by 
a wide range of nongenotoxic compounds. The common factor is 
prolonged stimulation of the thyroid by TSH following disruption of 
the normal feedback mechanism that controls the serum level of TSH. 
This disruption of thyroid hormone economy can be caused by 
interference with iodide uptake and thyroid hormone synthesis or 
secretion, interference with the peripheral metabolism of T4 or T3, 
or increased metabolism and excretion of thyroid hormones (see Refs. 
5 and 6).
    \6\ ``Ample information in experimental animals indicates a 
relationship between inhibition of thyroid-pituitary homeostasis and 
the development of thyroid follicular cell neoplasms. This is 
generally the case when there are long-term reductions in 
circulating thyroid hormones which have triggered increases in 
circulating thyroid stimulating hormone * * *. The progression of 
events leading to thyroid * * * neoplasms can be reversed under 
certain circumstances by reestablishing thyroid-pituitary 
homeostasis'' (Ref. 6).
---------------------------------------------------------------------------

    In further support of its argument, Hoechst submitted a set of 
publications addressing various aspects of thyroid function and 
toxicity, including thyroid carcinogenicity; a report authored by the 
``Acesulfame K Scientific Expert Panel,'' a group of experts retained 
by the petitioner to perform an independent safety evaluation of AAS 
and AAA (Ref. 7); and a letter from one of the experts from the 
Acesulfame K Scientific Expert Panel elaborating on the significance of 
the thyroid effects of AAA (Ref. 8).
    The petitioner also submitted the results of a short-term study of 
AAA in rats (the ``preliminary mechanistic study''). In this study, 5 
male rats per group were fed diets containing 0, 50, 123, 410, 1,110, 
or 2,400 ppm AAA or 90 ppm methimazole (positive control) for a period 
of 14 days. The following AAA-induced thyroid effects were observed in 
the preliminary mechanistic study: (1) Significantly increased absolute 
and relative thyroid weights in all positive control rats and in all 
rats fed diets containing 1,110 or 2,400 ppm AAA; (2) grossly enlarged 
thyroids in all positive control rats and in all rats fed diets 
containing 1,110 or 2,400 ppm AAA; (3) diffuse thyroid follicular cell 
hypertrophy and hyperplasia in all positive control rats and in all 
rats fed diets containing 1,110 or 2,400 ppm AAA; (4) significantly 
increased levels of TSH in positive control rats, as well as in rats 
fed 410, 1,110 or 2,400 ppm

[[Page 36350]]

AAA; (5) significantly decreased levels of T4 and reverse T3 in 
positive control rats and in rats fed diets containing 1,110 or 2,400 
ppm AAA; and (6) significantly decreased T3 levels in positive control 
rats and in rats fed diets containing 2,400 ppm AAA (see Ref. 4).
    In further support of its proposed mechanism, Hoechst also 
submitted the results of an in vitro investigation of the action of AAA 
on canine thyroperoxidase. In this study, AAA was shown to inhibit 
enzyme activity in a dose-related manner; the AAA concentration at 
which 50 percent enzyme inhibition occurred was calculated by Hoechst 
to be 28.6 micromolar. Hoechst pointed to the consistency between the 
results of both the preliminary mechanistic study and the 
thyroperoxidase inhibition study as further evidence for the link it 
hypothesized between thyroperoxidase inhibition and the thyroid-related 
effects observed in the oral toxicity studies of AAA.
    Hoechst also argued that a substance acting through a TSH-dependent 
mechanism would be expected to show a threshold below which no 
excessive stimulation of thyroid follicular cells would occur. The 
petitioner acknowledged that it is difficult to actually determine 
thresholds for low-incidence effects because of the small numbers of 
animals ordinarily used in toxicity studies (see Ref. 8). However, 
Hoechst cited the results of the preliminary mechanistic study, the 
results of the in vitro thyroperoxidase inhibition study, and the 
results of the short-term and subchronic oral studies in rats, rabbits, 
and dogs as strong evidence of the existence of a threshold for AAA-
induced thyroid effects. The petitioner also pointed to the negative 
results of the genetic toxicity tests of AAA as further support for its 
argument that a threshold level should exist, below which 
administration of AAA would not induce thyroid tumors. That is, 
hypertrophy and hyperplasia and, by extension, possible progression to 
tumors, would occur only at AAA doses high enough to increase 
circulating levels of TSH, and not through a genotoxic mechanism.
    In summary, Hoechst proposed the following nongenotoxic or 
``secondary'' mechanism for the AAA-induced effects observed in the 
thyroids of several species: (1) At high doses, AAA acts to disrupt 
thyroid hormone economy by inhibiting thyroperoxidase activity and thus 
decreasing serum levels of T3 and T4; (2) the disruption in thyroid 
hormone economy results in hypersecretion of TSH by the pituitary; (3) 
the elevated blood levels of TSH, if sustained, result in hypertrophy 
and hyperplasia of the thyroid follicular cells and, eventually, 
thyroid tumors; and (4) that AAA does not act through a genotoxic 
mechanism to initiate a neoplastic process.
    Hoechst explicitly acknowledged that there was a distinct 
possibility that AAA, if tested in a 2-year rodent bioassay, would 
induce thyroid tumors. However, Hoechst also maintained that thyroid 
tumors would occur only as a result of chronic consumption of AAA in 
amounts high enough to induce excess TSH production. Hoechst argued 
that because AAA would be consumed only in extremely low amounts, well 
below any value they believed likely for the postulated threshold for 
stimulating excess TSH production, it would be appropriate to base an 
analysis of the potential health risk from AAA on a comparison between 
the NOEL's for certain thyroid endpoints and the anticipated low levels 
of intake (a ``safety factor'' or ``threshold concept'' approach). 
Hoechst concluded that because the NOEL's for AAA's thyroid effects 
exceeded its dietary exposure estimate by a factor of approximately 2 
million, there would be essentially no risk to human health from 
dietary exposure to AAA resulting from consumption of beverages 
sweetened with ACK.
    FDA agrees that the anticipated human dietary exposure to AAA is 
lower than the NOEL's for AAA-related thyroid hypertrophy and 
hyperplasia by several orders of magnitude. FDA does not agree, 
however, that Hoechst's approach of simply comparing these NOEL's with 
dietary exposure is sufficient for evaluating the potential health risk 
suggested by the AAA-related effects observed in the thyroid. As 
previously noted, the AAA-related histopathological findings in the 
thyroid (i.e., hypertrophy and hyperplasia in rats, rabbits, and dogs, 
together with adenomas in two AAA-treated male rats in the subchronic 
study) suggest that AAA may induce thyroid tumors in long-term studies. 
Hoechst's ``safety factor'' approach relies on the firm's proposed 
mechanism for AAA's action on the thyroid, which explicitly 
incorporates a presumed threshold for AAA's thyroid effects. FDA has 
concluded, however, that the available data do not establish the 
mechanism proposed by the petitioner. The strengths and weaknesses in 
the data submitted in support of Hoechst's proposed mechanism are 
discussed in the following paragraphs.
    FDA has determined that there is strong evidence that AAA is not 
genotoxic. The agency also acknowledges that some of the results from 
the preliminary mechanistic study and the in vitro study of canine 
thyroperoxidase are consistent with Hoechst's argument that AAA-induced 
effects on the thyroid are mediated through disruption of thyroid 
hormone economy. In particular, because inhibition of thyroperoxidase 
would cause TSH serum levels to increase rapidly, the results of the in 
vitro thyroperoxidase inhibition study are consistent with results of 
the preliminary mechanistic study. The preliminary mechanistic study 
also provides some support for the hypothesis that AAA-induced thyroid 
effects in rats are mediated by dose-related perturbations in thyroid 
hormone economy because decreased circulating levels of T3 and T4 and 
increased serum TSH levels were associated with thyroid follicular cell 
hypertrophy and hyperplasia in this study.
    However, a threshold level for thyroperoxidase inhibition in vivo 
cannot be determined from the available data, which were obtained in an 
in vitro system. In addition, a threshold level for AAA-induced TSH 
induction cannot be determined from the in vivo studies, which were 
conducted with too few animals. Finally, the in vivo studies of AAA-
induced effects on thyroid hormone economy (the preliminary mechanistic 
study in rats and the second short-term dog study) were both limited to 
14 days duration; there are no studies of the effects of longer periods 
of exposure to AAA on thyroid hormone economy.
    Moreover, FDA has determined that some of the data from the short-
term and subchronic toxicity studies appear to be inconsistent with 
Hoechst's proposed mechanism. For example, as discussed above, early 
AAA-related changes in the thyroid (e.g., hypertrophy and hyperplasia), 
if induced via the petitioner's proposed mechanism, would be expected 
to be reversible. However, in the second 14-day dog study, one of the 
two high-dose animals with thyroid follicular hyperplasia was a 
``recovery'' animal (i.e., an animal sacrificed 1 month after dosing 
ended); the observation of hyperplasia in a ``recovery'' animal 
indicates that AAA's effect on the thyroid persisted for 1 month after 
dosing ended. This raises the possibility that the effect may persist 
for longer than 1 month and may not be readily or completely 
reversible.
    Similarly, some of the data obtained from the subchronic rat study 
are not entirely consistent with certain features of the mechanism 
proposed by Hoechst.

[[Page 36351]]

 Hoechst has advanced, as part of its argument, the observation that 
rodents are more susceptible to TSH-mediated thyroid effects than other 
species, and that male rats are ``particularly vulnerable.'' However, 
FDA notes that the available data do not show clear differences, 
between rats and dogs, in sensitivity to AAA-induced effects. For 
example, the NOEL for AAA-induced thyroid effects in rats in the 
subchronic study and the level at which no AAA-induced effects were 
observed in the second dog study are approximately the same. In 
addition, although FDA's review of the subchronic rat study showed that 
male rats may have been slightly more susceptible to AAA's thyroid 
effects than female rats, the differences were again small.
    FDA concludes that, for several reasons, the petitioner's proposed 
mechanism has not been established. First, as noted, some of the 
results of the short-term and subchronic feeding studies (e.g., 
persistence of thyroid effects in recovery animal in the dog study; the 
lack of a clear difference, in sensitivity to AAA, between rats and 
dogs and between male and female rats) appear to be inconsistent with 
the proposed mechanism. Second, the data on AAA's effects on thyroid 
hormone economy are limited to short-term exposures of a relatively 
small number of animals; as previously noted, these limited data do not 
permit the determination of a threshold for AAA's effects. Thus, FDA 
has determined that although the mechanism proposed by Hoechst is 
plausible, it has not been established. Because Hoechst's approach to 
evaluating the health risk from AAA (a comparison of the NOEL's for 
certain thyroid endpoints with dietary AAA exposure) relies explicitly 
on the firm's proposed mechanism, and the proposed mechanism has not 
been established, FDA concludes that Hoechst's approach is not 
sufficient for an evaluation of the health risk from AAA.
    vi. Consideration of whether more testing of AAA is necessary--(1) 
Statement of the issue. Because the findings in the short-term and 
subchronic toxicity studies of AAA suggest that AAA could induce 
thyroid tumors in a long-term study, FDA carefully considered whether 
conduct of such a study was necessary to evaluate the safety of ACK for 
use in nonalcoholic beverages. In particular, given the likely human 
dietary exposure to AAA, FDA considered whether the possibility that 
AAA might induce tumors in a long-term bioassay raised sufficient 
concern such that testing of the hypothesis should be required. Said 
differently, the issue was whether a long-term oral study of AAA, a 
hydrolysis product expected to be present at extremely low levels (if 
at all) in only certain nonalcoholic beverages, is needed to evaluate 
the safety of the petitioned use of the food additive, ACK. In 
addressing this question, FDA determined that it was critical to assess 
both the likely putative tumorigenic (neoplastic) potency of AAA and 
the likely patterns of dietary exposure to AAA resulting from 
consumption of ACK-sweetened nonalcoholic beverages.
    As discussed in detail in the rest of this section, FDA considered 
several approaches to assessing the risk from AAA, and determined both 
that long-term testing of AAA is unnecessary and that the petitioned 
use of ACK in nonalcoholic beverages is safe.
    (2) Risk assessment. The usual process of quantitative risk 
assessment is characterized by four steps. First, a possible 
toxicological hazard is identified. Second, mathematical modelling 
techniques are applied to the dose-response information from a toxicity 
study in order to estimate the probability, or, usually, an upper-bound 
limit on the probability, of the toxic effect of the substance at any 
given dose level (see for example, Refs. 9 through 11).\7\ Typically, 
in a risk assessment of a carcinogen, this dose-response information is 
taken from tumor incidence data from a long-term animal study; most 
often, this long-term study is conducted in a rodent species. Third, 
the likely human dietary exposure to the substance is estimated. This 
estimate of dietary exposure may consider such factors as the age 
groups likely to be exposed and the type, magnitude, and duration of 
the anticipated exposures.\8\ Finally, the information from the first 
three steps is combined to characterize the risk associated with the 
potential human exposure to the substance in question.
---------------------------------------------------------------------------

    \7\ In the absence of information that would support another 
approach, FDA uses simple linear extrapolation from the dose-
response information in the experimental range to estimate the dose-
response outside the experimental range (that is, at lower doses 
comparable to the anticipated human exposure).
    \8\ In the risk assessment of carcinogenic constituents of food 
and color additives used directly in food, FDA most often uses an 
estimate of the lifetime-averaged daily dietary exposure to the 
substance in question.
---------------------------------------------------------------------------

    In the present case, as in the usual risk assessment process, a 
possible hazard, thyroid carcinogenicity, has been identified. There 
are similarities between the thyroid effects produced by oral 
administration of AAA in short-term and subchronic toxicity studies and 
those produced by oral administration of other substances known to 
induce thyroid tumors in long-term rodent studies. Thus, there is the 
possibility that AAA would also induce tumors if tested in a long-term 
rodent study and, thus, may ultimately present a carcinogenic hazard to 
humans.
    The risk assessment process used in the present case differs from 
the usual process, however, in that AAA has not been demonstrated to be 
an animal (or human) carcinogen. That is, dose-response information 
from a long-term oral study of AAA in animals has not been used because 
such a study has not been conducted. As an alternative, FDA has used 
information from the many existing long-term oral studies of known 
thyroid tumorigens to assess the probable carcinogenic potency (or 
range of probable potencies) of AAA that might be determined, were a 
carcinogenicity study of AAA conducted in a rodent species. The agency 
believes this is a sound approach because of the substantial amount of 
information available for a large number of thyroid tumorigens.\9\
---------------------------------------------------------------------------

    \9\ Potency values at the thyroid and at other organ sites are 
available for a large number of thyroid tumorigens. In addition, the 
results of genetic toxicity testing, short-term studies, and other 
toxicity testing are available for many of the these compounds. 
Mechanistic information, though not complete in many cases, is also 
available for a significant number of these compounds, as well as 
information regarding structure-activity relationships.
---------------------------------------------------------------------------

    As in the usual risk assessment process for a known carcinogenic 
constituent of a food or color additive, a potential life-time averaged 
``daily'' human dietary exposure to the substance in question (in this 
case, AAA, a putative tumorigen) has been estimated. In calculating 
this estimate, FDA has used estimates of the likely human dietary 
exposure to ACK, in conjunction with information from analytical 
testing conducted on model solutions under exaggerated conditions, to 
estimate a potential lifetime-averaged level of daily dietary exposure 
to AAA. FDA's exposure estimate is conservative in that it incorporates 
numerous assumptions and default values for certain parameters that, 
when combined, yield a value for ``daily'' dietary exposure to AAA that 
is likely to overestimate rather than underestimate such exposure. By 
combining the information regarding potential human dietary exposure 
with the information regarding the likely tumorigenic potency (or range 
of probable potencies) of AAA, FDA has characterized the potential 
human carcinogenic risk from AAA resulting

[[Page 36352]]

from the consumption of ACK-sweetened nonalcoholic beverages.
    The petitioner and the agency have separately analyzed the likely 
health risk suggested by the AAA-related thyroid findings in the short-
term studies, by considering both estimates of the tumorigenic potency 
of AAA and the likely patterns of dietary exposure to AAA resulting 
from consumption of ACK-sweetened nonalcoholic beverages. In the course 
of its analysis, scientists from FDA's Center for Food Safety and 
Applied Nutrition consulted with several scientists (hereafter referred 
to as ``the FDA consultants''), from both within and outside the 
agency, with expertise in various scientific disciplines relevant to 
the agency's analysis. Details of the petitioner's analysis and the 
agency's analysis (including relevant comments from the FDA 
consultants) are discussed in the following paragraphs.
    (3) Hoechst's analysis. In response to the agency's reservations 
regarding Hoechst's initial, threshold-based approach to evaluating the 
potential health risk from AAA, Hoechst performed two additional 
``extreme-case'' or ``worst-case'' comparative risk assessments. In 
both assessments, Hoechst assumed that AAA would induce thyroid tumors 
in a long-term study, even though AAA has not been shown to be a 
tumorigen. In contrast to the firm's initial approach, neither of 
Hoechst's comparative risk assessments was predicated on a threshold 
for AAA's thyroid effects. That is, both of Hoechst's comparative risk 
assessments assumed that some risk of neoplastic disease would be 
present at all levels of exposure to AAA.
    In presenting its assessments of the tumorigenic potential of AAA, 
Hoechst continued to argue strongly for the mechanism it had proposed 
to account for AAA's thyroid effects. Hoechst used several features of 
its proposed mechanism to select the set of chemicals against which to 
compare AAA and estimate AAA's tumorigenic potential; Hoechst's 
selection of these surrogates for AAA is described in the following 
paragraphs.
    Using data from lifetime studies of thyroid tumorigens that Hoechst 
identified as acting with similar effect and through a mechanism 
similar to the one it had proposed for AAA, Hoechst estimated AAA's 
putative thyroid tumor potency. According to Hoechst, these estimates 
of AAA's putative thyroid tumor potency, coupled with an estimate of 
dietary exposure, would provide ``comparative risk assessments'' of 
AAA's potential to induce thyroid tumors. Hoechst drew upon several 
recognized sources to identify the thyroid tumorigens that it chose as 
surrogates for AAA. These sources included a publication analyzing 
target organs for more than 500 chemicals in the Carcinogen Potency 
Database (CPDB), a published review of the information in the data base 
maintained by the National Toxicology Program (NTP), the Integrated 
Risk Information System (IRIS), and a well known literature source on 
thyroid follicular cell carcinogenesis (Refs. 6 and 12 through 14).\10\ 
From the group of thyroid tumorigens identified using these sources, 
Hoechst selected those for which long-term rodent bioassays had been 
conducted and in which the test substance displayed tumorigenic 
activity in either the thyroid alone or, if tumorigenic at other organ 
sites as well, with greater potency at the thyroid than at other sites. 
From this subset of thyroid tumorigens, only those compounds that 
Hoechst identified as both nonmutagenic and active in inhibiting 
thyroperoxidase (both of which are critical elements of Hoechst's 
proposed mechanism) were retained as AAA surrogates. Applying these 
criteria, Hoechst identified four compounds: Amitrole, methimazole, 
propylthiouracil, and sulfamethazine.
---------------------------------------------------------------------------

    \10\ The CPDB summarizes results of carcinogenicity bioassays 
published in the open literature and in technical reports of the 
NTP. The NTP data base, also known as the NCI/NTP data base, 
contains the results of mouse and rat carcinogenicity studies 
conducted by NCI/NTP. The published review that was used by Hoechst 
summarized the results of 343 selected carcinogenicity studies 
conducted by NCI/NTP; in this subset of the NCI/NTP data base, 14 
percent of the studies in male rats, 11 percent of the studies in 
female rats, 8 percent of the studies in male mice and 9 percent of 
the studies in female mice were identified as having positive or 
equivocal, chemically-related thyroid proliferative lesions. (The 
studies from the NCI/NTP data base are also included in the CPDB.) 
IRIS is an electronic data base prepared and maintained by the U.S. 
Environmental Protection Agency (EPA); it contains information on 
human health effects that may result from exposure to various 
chemicals in the environment.
---------------------------------------------------------------------------

    Hoechst used the same estimated dietary exposure in both of its 
comparative risk assessments. In calculating this estimate, Hoechst 
used data on ACK stability and nonalcoholic beverage consumption 
patterns, incorporating several conservative assumptions similar to 
those used by FDA and described previously. Hoechst estimated the high-
level consumer's potential ``daily'' dietary exposure to AAA to be 3.5 
ng/kg bw/day. Hoechst asserted that this estimate of potential 
``daily'' dietary exposure was likely to overestimate significantly the 
actual exposure because of the numerous conservative assumptions used 
in deriving the estimate.\11\
---------------------------------------------------------------------------

    \11\ Hoechst's estimate of consumer exposure to AAA (3.5 ng/kg 
bw/d) is essentially the same as FDA's estimate (3.3 ng/kg bw/d, 
equivalent to 0.2 g/p/d). FDA has determined that both 
Hoechst's and the agency's estimate of AAA dietary exposure, because 
of the particular assumptions used in deriving them, are likely to 
overestimate rather than underestimate exposure.
---------------------------------------------------------------------------

    In its first comparative risk assessment, Hoechst assumed that the 
putative induction of thyroid tumors by AAA would be directly related 
to an AAA-induced increase in serum levels of TSH. Using the literature 
sources listed previously, Hoechst identified three compounds 
(methimazole, propylthiouracil, and sulfamethazine) that the firm 
asserted have approximately the same quantitative effect on circulating 
TSH levels as AAA had on TSH levels in the preliminary mechanistic 
study in rats. Hoechst then estimated a hypothetical cancer potency for 
AAA by interpolating between the established tumorigenic potencies of 
these three substances;\12\ the hypothetical cancer potency for AAA in 
this assessment was 2.3 x 10-3 (mg/kg bw/day)-1. 
When coupled with the firm's estimated ``daily'' dietary exposure of 
3.5 ng/kg bw/day, Hoechst's estimated upper-bound limit of lifetime 
human cancer risk, in its first assessment, was 8.1 x 10-9.
---------------------------------------------------------------------------

    \12\ The potencies of the AAA surrogates are properly described 
as tumorigenic potencies; the tumors observed in rodents are more 
often benign, rather than malignant, follicular cell tumors. In both 
the petitioner's and the agency's comparative risk assessments, the 
distribution of tumorigenic potencies of AAA surrogates is used to 
estimate the putative tumorigenic potency of AAA. This putative 
tumorigenic potency of AAA is then used as a direct substitute for a 
hypothetical human cancer potency in the comparative risk 
assessments.
---------------------------------------------------------------------------

    In the second of Hoechst's nonthreshold risk assessments, the 
putative induction of thyroid tumors by AAA was assumed to be directly 
related to AAA-induced inhibition of thyroperoxidase (and thus, 
indirectly, to elevated serum TSH levels). Hoechst identified four 
substances (amitrole, methimazole, propylthiouracil, and 
sulfamethazine) for which it maintained that the induction of thyroid 
tumors in animals is known to occur as a result of thyroperoxidase 
inhibition. Hoechst then estimated a hypothetical cancer potency for 
AAA by calculating a weighted average of the established tumorigenic 
potencies of these four substances. In this second comparative risk 
assessment, Hoechst estimated the hypothetical potency of AAA as 4.0 x 
10-2 (mg/kg bw/day)-1. When coupled with the 
firm's estimated ``daily'' dietary exposure of 3.5 ng/kg bw/day, 
Hoechst's estimated upper-bound limit

[[Page 36353]]

of lifetime human cancer risk, in its second assessment, was 
approximately 1.4 x 10-7.
    The petitioner argued that both its estimates of AAA's upper-bound 
limit of lifetime human cancer risk were well below the level 
ordinarily regarded by FDA as commensurate with negligible risk. The 
petitioner also argued that any actual risk would be far lower than 
these estimated upper-bound limits of risk because of the numerous 
conservative assumptions used in calculating these estimates.
    In addition, the petitioner noted that humans are less sensitive 
than rats to thyroid effects induced through TSH-dependent mechanisms. 
Hoechst referenced scientific literature in support of its contention 
that, although chronic TSH stimulation induces thyroid hypertrophy and 
hyperplasia in humans as well as in rodents, humans are less likely to 
develop tumors following chronic stimulation by TSH. Specifically, they 
noted that prolonged TSH stimulation is known to lead to thyroid 
enlargement or goiter in humans, but rarely leads to thyroid tumors 
(Refs. 15 and 16). Hoechst also maintained that the rat's significantly 
higher baseline TSH levels and more rapid metabolism of the hormone 
leave rats more vulnerable than humans to the development of thyroid 
tumors in response to chemically induced increases in circulating TSH 
levels (see Refs. 8 and 17). Hoechst argued that the lower sensitivity 
of human thyroid follicular cells to elevated TSH levels would further 
reduce the likely magnitude of any actual thyroid tumor risk to humans 
from exposure to any AAA in ACK-sweetened nonalcoholic beverages.
    (4) FDA's analysis. FDA has carefully evaluated the petitioner's 
comparative risk assessments. The agency agrees that it is reasonable 
to perform an ``extreme-case'' risk assessment of AAA in order to 
evaluate the potential health concern raised by the thyroid findings in 
the short-term studies of AAA. To this end, FDA conducted its own 
analysis of the potential health risk from the low levels of AAA that 
may be ingested as a result of the consumption of ACK-sweetened 
nonalcoholic beverages. FDA's two principal comparative risk 
assessments of AAA, like the petitioner's, are essentially modified 
carcinogenic risk assessments; however, in several respects the 
agency's approach differs from the petitioner's.
    Like Hoechst, FDA assumed that AAA would be tumorigenic if tested 
in a long-term bioassay. The agency also assumed, as did Hoechst in its 
comparative risk assessments, that there is no threshold for AAA's 
presumed tumorigenic activity. However, in contrast to Hoechst, FDA did 
not rely on assumptions regarding AAA's mechanism of action on the 
thyroid. Although FDA believes that it is plausible that AAA may induce 
thyroid tumors in long-term studies through the mechanism hypothesized 
by the petitioner, the data supporting the petitioner's hypothesis are 
limited in several key areas. First, as noted, there are no studies 
demonstrating long-term effects of AAA on thyroid hormone economy; 
thus, FDA, in its comparative risk assessments, did not assume a 
quantitative correlation between TSH induction and AAA's putative 
thyroid tumorigenic potency. Second, there is no direct evidence of 
AAA-induced effects on thyroperoxidase activity in vivo; consequently, 
FDA did not assume that AAA's putative potency would be similar to 
potencies of thyroid carcinogens known or asserted to act through 
inhibition of thyroperoxidase activity.
    To provide assurance that the risk presented by AAA is not 
underestimated, FDA included in its set of AAA surrogates all 
substances it identified, using the 1996 CPDB (see Ref. 18), as having 
induced tumors in the thyroid, including substances that also induced 
tumors in other organs, regardless of the relative potencies 
involved.\13\ This set of surrogates includes both genotoxic and 
nongenotoxic substances. Because the potency distribution for genotoxic 
chemicals is shifted to higher potencies than the potency distribution 
for nongenotoxic chemicals, FDA's set of 91 surrogates includes 
substances of higher potency than those in Hoechst's set of 4 
surrogates (Ref. 2). FDA included this frank and deliberate 
conservatism to ensure that neither the putative potency of AAA nor the 
attendant estimate of AAA's potential carcinogenic risk would be 
underestimated.
---------------------------------------------------------------------------

    \13\ Taken together, the six plots of the 1996 CPDB include 
results of 5,002 experiments on 1,230 chemicals. The agency notes 
that of the 91 compounds in the CPDB that were reported to induce 
thyroid tumors in rodents, only three (methimazole, deltamethrin, 
and sulfamethazine) produced thyroid tumors only. Of the remaining 
88 compounds, 70 percent had a higher cancer potency for tumors 
other than thyroid tumors. Thus, the majority of compounds that have 
been found to induce thyroid tumors (by any mechanism) have also 
been found to induce tumors at other sites, for which the estimated 
cancer potency is higher than the potency estimated for thyroid 
tumors alone (see Ref. 2).
---------------------------------------------------------------------------

    In the first of FDA's comparative risk assessments, the agency used 
potency values from the distribution of the thyroid tumor potencies of 
the 91 surrogates. FDA chose this approach because the data from the 
short-term and subchronic studies of AAA in rats, rabbits, and dogs 
identify the thyroid as the potential target organ for putative AAA-
induced tumors and do not suggest other likely target organs. The 
distribution of thyroid tumor potencies for the 91 surrogates has a 
peak, or ``most probable'' value, of 7.0 x 10-3 (mg/kg bw/
day)-1. FDA used this potency value as an estimate for the 
likely potency of AAA. This potency, coupled with the agency's 
estimated ``daily'' dietary exposure to AAA of 3.3 ng/kg bw/day, yields 
an estimated upper-bound limit of lifetime risk from AAA of 2.3 x 
10-8 (Ref. 2). This hypothetical upper-bound limit of 
lifetime risk from AAA is well below the level that FDA ordinarily 
considers commensurate with negligible risk.
    To provide further assurance that AAA's potential risk was not 
being underestimated, the agency performed a second risk assessment. In 
this second assessment, FDA hypothesized that AAA might, in addition to 
inducing thyroid tumors, induce tumors at sites other than the thyroid 
and that AAA's potency at these other sites could be higher than for 
tumors induced at the thyroid.\14\ In essence, this scenario describes 
the most adverse outcome of a long-term bioassay with AAA, were such a 
bioassay actually conducted. Thus, FDA's second risk assessment 
included an assumption of the most adverse outcome for a study testing 
the hypothesis that AAA causes thyroid tumors so that the potential 
risk posed by AAA would not be underestimated.
---------------------------------------------------------------------------

    \14\ One of the FDA consultants noted that some, but not all 
thyroid peroxidase inhibitors lead to tumors at sites other than the 
thyroid, especially the liver of mice. This consultant further 
commented that ``* * * FDA is on strong ground to look at the 
potency for tumors other than thyroid, as well as looking at those 
for the thyroid.'' Including the higher potencies for tumors other 
than thyroid tumors in FDA's assessment is, however, a conservative 
measure in that the data in the studies of AAA submitted to the 
petition do not suggest that there are other likely target organs 
for neoplasia.
---------------------------------------------------------------------------

    In this assessment, to estimate AAA's most likely tumorigenic 
potency, FDA used the peak, or ``most probable value'' value from the 
distribution of highest tumor potencies at any organ site for FDA's 91 
surrogates. Using this estimate of the putative tumorigenic potency of 
AAA (2.0 x 10-2 (mg/kg bw/d)-1) and the agency's 
conservative estimate of ``daily'' dietary exposure to AAA of 3.3 ng/kg 
bw/d, FDA estimated the upper-bound limit of lifetime human cancer risk 
from exposure to AAA to be 6.6 x 10-8 (Ref. 2). This 
hypothetical upper-bound limit of lifetime risk from AAA, like the 
value obtained in FDA's first

[[Page 36354]]

risk assessment, is well below the level ordinarily considered by FDA 
as commensurate with negligible risk.
    Based on its risk assessments, the agency believes that AAA is 
highly unlikely to pose more than a negligible cancer risk to 
consumers. For example, even if, in FDA's first risk assessment, AAA's 
thyroid tumor potency were as high as that of the 90th percentile most 
potent compound in FDA's set of AAA surrogates, the estimated upper-
bound limit of lifetime risk from AAA, using all of the conservative 
features and assumptions described previously, would still be less than 
7 x 10-7. To produce the same estimate of upper-bound risk 
from AAA using the approach in FDA's second risk assessment, AAA's 
potency at any organ site would have to approach that of the 90th 
percentile most potent compound in FDA's set of AAA surrogates. The 
agency considers these potency levels highly unlikely for several 
reasons. First, AAA's potency at the thyroid would need to approach 
that of methimazole, the positive control in the preliminary 
mechanistic study. That AAA would be as potent as methimazole is 
unlikely, however, given the fact that almost 100-fold greater doses of 
AAA than of methimazole were needed to induce comparable degrees of 
thyroid follicular cell hypertrophy and hyperplasia, the presumed 
precursors to any thyroid neoplasia (see Ref. 2). Second, the thyroid 
tumorigens in the set of 91 surrogates with potencies in this range 
(approaching the 90th percentile and above) are almost all genotoxic or 
have strong structural indicators of genotoxicity while the results of 
the genetic toxicity tests of AAA show that AAA is not genotoxic. As 
previously noted, the potency distribution for genotoxic compounds is 
shifted to higher values than the potency distribution of nongenotoxic 
compounds; thus, the probability that AAA, a nongenotoxic compound, 
will be more potent than the most potent genotoxic compounds in FDA's 
set of AAA surrogates is extremely low (see Ref. 2).
    As noted previously, the agency's comparative risk assessments were 
based on numerous conservative assumptions so that any risk from AAA 
would not be underestimated; FDA believes that any actual risk from AAA 
would be substantially lower than either of its estimates of the upper-
bound limit of lifetime risk. The agency also notes that all of the FDA 
consultants agreed that the numerous conservative assumptions used in 
the agency's comparative risk assessments were likely to lead to an 
overestimate, rather than an underestimate, of the risk from AAA.\15\
---------------------------------------------------------------------------

    \15\ One of the FDA consultants also provided two additional 
approaches to calculating a conservative upper-bound limit of 
lifetime human cancer risk, one that made use of a feature of the 
petitioner's proposed mechanism for AAA's action on the thyroid and 
one that did not. The estimates of AAA's upper-bound carcinogenic 
risk derived by these two additional approaches were 8.0 x 
10-8 and 3.3 x 10-8, respectively (see Ref. 
2). Both of the consultant's estimates for the upper-bound risk from 
AAA, like the upper-bound risks calculated by FDA (2.3 x 
10-8 and 6.6 x 10-8) and by the petitioner 
(8.1 x 10-9 and 1.4 x 10-7), are very low.
---------------------------------------------------------------------------

    The conservative nature of FDA's risk estimates was amplified by 
the agency's assumption, in its comparative risk assessments, that 
consumers would be subject to ``chronic'' or ``daily'' dietary exposure 
to AAA through consumption of ACK-sweetened nonalcoholic beverages. In 
fact, frequent exposure to AAA is unlikely because few containers of 
beverages are likely to be stored under the conditions necessary to 
produce significant quantities of AAA. Thus, any actual dietary 
exposure to AAA through consumption of ACK-sweetened beverages is 
likely to be at very low levels, to be intermittent, and to be 
infrequent.\16\
---------------------------------------------------------------------------

    \16\ FDA notes that approaches to modifying risk assessments for 
intermittent exposures to carcinogens generally reduce the estimated 
risk substantially (see for example, Refs. 19 and 20). Such 
modification can be particularly important for carcinogens that are 
nongenotoxic. In general, continuous exposure to such substances for 
a prolonged period of time is needed before tumors develop; removal 
of the carcinogen from the diet for a significant portion of that 
time, will stop progression toward tumor development and may even 
result in partial or complete reversal of the treatment-related 
preneoplastic changes (see Ref. 6). If AAA were to induce thyroid 
tumors, and if it were to do so through a nongenotoxic or indirect 
mechanism, the intermittent nature of the exposure to AAA from 
consumption of ACK-sweetened nonalcoholic beverages would reduce the 
risk from AAA so that it is even more likely to be significantly 
less than the value estimated by the agency's method, and perhaps to 
be zero. On this point, one of the FDA consultants also commented 
that explicit consideration of the expected intermittent nature of 
any dietary exposure to AAA was particularly important in placing 
the calculations of AAA's estimated risk into perspective.
---------------------------------------------------------------------------

    In summary, the agency has used information from the many long-term 
oral studies of known thyroid tumorigens to estimate the range of 
possible tumorigenic potencies of AAA; this estimate has then been used 
to represent the tumorigenic potency for AAA that might be determined 
by a carcinogenicity study of AAA in a rodent species. FDA has combined 
this information with a conservative estimate of ``daily'' dietary 
exposure to AAA in order to assess the risk that might be posed to 
individuals consuming ACK-sweetened beverages. FDA's risk assessments 
for AAA all yield upper-bound limits of lifetime risk that are not only 
very low, but are also expected to be substantially higher than any 
actual risk from AAA.
    (5) Resolution of the issue. FDA has carefully evaluated the data 
from the available short-term and subchronic oral toxicity tests of 
AAA. As previously noted, the findings in these studies suggested that 
AAA might induce thyroid tumors in a long-term oral study, raising the 
question of AAA's possible carcinogenic risk. Thus, FDA has considered 
whether conduct of a long-term study was necessary to assess the 
possible carcinogenic risk from AAA.
    FDA has concluded that, for several reasons, it is not necessary to 
require the conduct of a long-term study of AAA. First, the primary 
purpose of such a study would be to determine whether AAA actually 
induced thyroid tumors. As an alternative, in its assessment of the 
potential health risk of AAA, the agency has simply chosen to assume 
that AAA would, indeed, induce thyroid tumors in a long-term study, 
thus obviating the first purpose of such a study.
    The second purpose of a long-term study of AAA, in the event that 
AAA were found to be tumorigenic, would be to determine AAA's 
tumorigenic potency. As an alternative, in its risk assessments for 
AAA, FDA has conservatively estimated AAA's putative potency by 
considering the range of potencies of the many known thyroid tumorigens 
(AAA surrogates) for which long-term testing has been conducted. As 
noted previously, FDA believes this is a sound approach because the 
results of the short-term tests of AAA indicate the thyroid as a likely 
target organ for the assumed neoplasia, and because of the substantial 
amount of chemical and toxicological information available for a large 
number of thyroid tumorigens.
    FDA has also used several deliberate conservatisms in constructing 
its set of surrogates in order to ensure that AAA's putative potency 
and any attendant estimate of AAA's hypothetical cancer risk are not 
underestimated: (1) FDA's set of surrogates includes genotoxic 
compounds which, as a group, are generally more potent than 
nongenotoxic compounds (AAA is nongenotoxic); (2) FDA's set of AAA 
surrogates also includes compounds for which genetic toxicity testing 
data are not available, but which have features in their chemical 
structures that are widely recognized as strong indicators of 
mutagenicity/carcinogenicity and, thus, are expected to be of higher 
potency than nongenotoxic compounds; and (3)

[[Page 36355]]

FDA's set of surrogates includes thyroid tumorigens that are 
tumorigenic at sites other than the thyroid and with higher potency 
than at the thyroid. Using information regarding the AAA surrogates and 
the distribution of their potencies, FDA estimated a range of 
hypothetical carcinogenic potencies for AAA. Thus, by conservatively 
estimating the range of likely tumorigenic potencies for AAA, FDA 
believes that it has obviated the need to determine AAA's potency 
through long-term testing.
    Using the estimates of AAA's likely tumorigenic potency, the agency 
performed several comparative risk assessments for AAA, combining the 
estimates of AAA's potency with a deliberately exaggerated estimate of 
dietary exposure to AAA to assess the possible risk from the compound; 
these conservative estimates of AAA's hypothetical upper-bound limit of 
cancer risk are very low. As previously noted, the risk estimates 
calculated by the FDA consultant and by Hoechst, though derived using 
different assumptions about the range of possible potencies for AAA, 
are also very low. In addition, the conservative nature of all of the 
risk estimates for AAA is amplified by the assumption that consumers 
would be subject to ``chronic'' or ``daily'' exposure to AAA through 
consumption of ACK-sweetened nonalcoholic beverages when, in fact, such 
exposure is likely to be both intermittent and infrequent.
    FDA's risk assessments show that, even assuming that AAA were 
carcinogenic in a long-term test, the hypothetical upper-bound of risk 
associated with an exaggerated estimate of dietary exposure to the 
compound would be extremely small. Because of the numerous 
conservatisms used in calculating these upper-bound limits of risk, FDA 
concludes that any actual risk from AAA would be far lower than these 
limits and, in fact, negligible. In this way, the results of FDA's risk 
assessments corroborate the agency's determination that a long-term 
study of AAA is not necessary to assess the potential risk to the 
public health from consumption of this compound.
    Thus, based on the available data and information, including the 
risk assessments described previously, FDA concludes that there is a 
reasonable certainty that no harm will result from the exposure to AAA 
that might result from the proposed use of ACK in nonalcoholic 
beverages. Accordingly, the agency has determined that requiring the 
petitioner to conduct further testing of AAA is not necessary and would 
not serve a useful purpose from the public health perspective.

E. Summary of FDA's Safety Evaluation

    The safety of ACK has been thoroughly tested and the data have been 
carefully reviewed by the agency. FDA has considered the data and 
information submitted in the present petition as well as other 
information in its files, including data and information in previous 
petitions for ACK.
    The agency has determined that the toxicological data on ACK 
establish that: (1) There is no association between neoplastic disease 
(cancer) and consumption of the additive and (2) the ADI for the 
additive is 15 mg/kg bw/day. FDA has also determined that the estimated 
dietary exposure to ACK from all currently permitted uses of the 
additive as well as the proposed use in nonalcoholic beverages (1.6 mg/
kg bw/day for the mean consumer, 3.0 mg/kg bw/day for the 90th 
percentile consumer) is well below the ADI. In addition, the agency has 
concluded that there is a reasonable certainty of no harm from the 
exposure to methylene chloride (a chemical used in the manufacture of 
ACK) that might result from all currently permitted uses of the 
additive as well as the proposed use in nonalcoholic beverages.
    Finally, FDA has considered the special conditions that are 
relevant to the proposed use in nonalcoholic beverages. In this regard, 
FDA has considered toxicological data and other information, including 
estimates of dietary exposure, regarding AAS and AAA, the principal 
hydrolysis products of ACK. Based on the data and information described 
previously in this document, including FDA's comparative risk 
assessments for AAA, the agency has concluded that there is a 
reasonable certainty of no harm from the exposure to AAS and AAA that 
might result from the proposed use of ACK in nonalcoholic beverages.
    Thus, based on a full and fair evaluation of the relevant data and 
information, FDA concludes that the proposed use of ACK in nonalcoholic 
beverages is safe.

IV. Response to Comments

    During the course of FDA's evaluation of the present petition, the 
agency received several sets of comments on the petition. FDA received 
multiple submissions from CSPI, who also transmitted comments from 
other interested parties. Later, Hoechst transmitted additional remarks 
from two of these same parties. Several letters were also received from 
trade groups and other organizations.

A. Summary of Comments

1. Center for Science in the Public Interest's (CSPI's) First 
Submission
    The first of CSPI's submissions was a letter, dated October 18, 
1990, in which CSPI referred to the organization's 1988 objections to 
FDA's initial approval of the use of ACK (the dry uses final rule). 
CSPI asked that FDA not consider expanding the permitted uses of ACK 
``without first resolving [CSPI's] objections, hearing request, and 
petition\17\ [sic].'' As noted previously in this document, FDA 
considered the issues raised by CSPI in its objections and responded, 
in detail, to those objections in the Federal Register of February 27, 
1992 (57 FR 6667). After reviewing the objections, the agency concluded 
that no genuine issues of material fact had been raised that would 
justify either a hearing or a stay of the regulation and, accordingly, 
denied CSPI's requests. Because the agency has responded to CSPI's 
objections to the dry uses final rule and to the organization's related 
requests, no further discussion of CSPI's first submission is 
warranted.
---------------------------------------------------------------------------

    \17\ CSPI uses the term ``petition'' to refer to its request for 
a stay of the dry uses final rule.
---------------------------------------------------------------------------

2. CSPI's Second Submission
    CSPI's second submission was a letter, dated January 29, 1996, in 
which CSPI asserted that the long-term toxicity testing of ACK was 
inadequate and that ACK was ``possibly carcinogenic.'' Once again, CSPI 
referred to its previous objections to the dry uses final rule, and 
urged FDA to deny the present petition and to require the petitioner to 
conduct additional carcinogenicity testing of ACK. CSPI did not, 
however, supply any substantive information to support these 
requests.\18\ In its letter, CSPI also mentioned certain results from 
the toxicity tests of AAA\19\ in support of its request for additional 
carcinogenicity testing of ACK, but did not supply any substantive 
information that had not already been considered by FDA or any 
explanation of how the AAA test results related to the organization's 
request for additional testing of ACK. Because CSPI did not provide any 
substantive information to support its requests, no

[[Page 36356]]

further discussion of this submission is warranted.
---------------------------------------------------------------------------

    \18\ In its January 29, 1996, letter, CSPI indicated that it 
intended to submit a detailed analysis of the ACK safety data at a 
future date.
    \19\ CSPI mentioned histologic changes in the thyroid glands of 
rats, rabbits, and dogs, referring specifically to ``hypertrophic 
and neoplastic changes'' when AAA was administered at high dose 
levels in short-term studies. As previously noted in this document, 
AAA-related thyroid follicular cell hypertrophy occurred in all 
three animal species; adenomas occurred only in two male rats in a 
subchronic study.
---------------------------------------------------------------------------

3. CSPI's Third Submission
    CSPI's third submission consisted of a letter to FDA, dated May 29, 
1996, in which CSPI reiterated its concerns about the carcinogenicity 
testing of ACK, and also included copies of the materials the 
organization had submitted to the National Toxicology Program (NTP) in 
nominating ACK for ``chronic toxicity (carcinogenicity) testing'' by 
NTP (``CSPI's NTP nomination package''). CSPI's NTP nomination package 
consisted of a cover letter, dated May 29, 1996, and a narrative 
describing CSPI's rationale for nominating ACK for testing under the 
NTP program (a document entitled ``Summary of Data on Acesulfame 
Potassium''), including a list of nine references and seven 
attachments.\20\
---------------------------------------------------------------------------

    \20\ FDA has assumed that the NTP nomination package is the 
detailed analysis of the safety data on ACK that CSPI indicated, in 
its letter of January 29, 1996, that it would send to the agency at 
a future date.
---------------------------------------------------------------------------

    The seven attachments in CSPI's NTP nomination package were three 
FDA review memoranda; the final report for a subchronic toxicity study 
of ACK in rats; a letter from Hoechst responding to FDA questions 
regarding histopathology data from two of the long-term studies of ACK 
in rodents; and two FDA memoranda, each summarizing a different meeting 
of Hoechst and FDA representatives. The agency notes that the 
attachments are all copies of publicly available documents contained in 
the administrative record for the dry uses final rule. The agency also 
notes, however, that CSPI did not provide NTP with all of the 
information from the administrative record for the dry uses final 
rule.\21\ Specifically, CSPI did not provide NTP with the reports on 
the long-term studies of ACK in rats or mice, the reports of the 
genetic toxicity studies of ACK, or any of the review memoranda from 
FDA's pathologists or FDA's Cancer Assessment Committee.
---------------------------------------------------------------------------

    \21\ The administrative record for the dry uses final rule 
contains all of the Hoechst study reports submitted in support of 
the original petition for ACK, other data and supporting 
information, FDA review memoranda, and other documents. Hoechst 
submitted reports for 6 genetic toxicity tests, 2 acute toxicity 
studies, a subchronic toxicity study, 4 reproduction or 
developmental toxicity studies, 3 long-term studies in rodents 
referred to previously in this document, a 2-year study in dogs, 11 
metabolism studies, and 7 other specialized studies.
---------------------------------------------------------------------------

    The narrative describing CSPI's rationale for nominating ACK for 
NTP testing raised various issues with respect to the three long-term 
ACK feeding studies in rodents that were submitted in the original ACK 
petition. FDA's analysis of the specific issues raised in CSPI's third 
submission is discussed in section IV.B.2 of this document.
4. CSPI's Fourth Submission
    CSPI's fourth submission consisted of a letter, dated July 31, 
1996, addressed to the Director of FDA's CFSAN, in which the 
organization reiterated its concerns regarding the long-term testing of 
ACK and also mentioned its nomination of ACK for chronic toxicity 
(carcinogenicity) testing by NTP. In addition, CSPI cited certain of 
the results from the toxicity testing of AAA and urged FDA to require 
the petitioner to conduct long-term testing of AAA. CSPI again asked 
FDA to deny the present petition and to revoke ``all existing 
regulations permitting the use of acesulfame potassium.''
    In support of its requests, CSPI enclosed copies of letters from 
``ten experts in the fields of carcinogenesis, toxicology, and 
statistics'' who had, at CSPI's request, ``reviewed the Hoechst test 
protocols and results'' (hereinafter, these individuals will be 
referred to as ``CSPI's ten consultants''). Seven of the letters were 
addressed to CSPI; the authors of these particular letters expressed 
support for CSPI's nomination of ACK for testing under the NTP program. 
Three of the letters were addressed to the Commissioner of the Food and 
Drug Administration. The authors of these three letters urged FDA to 
require additional carcinogenicity tests of ACK; one of the authors 
also urged FDA not to approve the present petition.\22\ CSPI claimed 
that ``[b]ased on the experts' conclusions regarding Hoechst's tests, 
it is clear that Hoechst has failed to demonstrate a 'reasonable 
certainty of no harm' for the use of acesulfame potassium in soft 
drinks (or other foods).''
---------------------------------------------------------------------------

    \22\ Several of the letters to CSPI and to FDA raised specific 
issues regarding the procedures used in, or the interpretation of 
results from, the long-term studies of ACK in rodents. None provided 
any new data or other information that had not already been 
considered by the agency. FDA's analysis of the specific issues 
raised in these letters is discussed later in this document.
---------------------------------------------------------------------------

    In partial response to CSPI's letter of July 31, 1996, FDA 
requested copies of the materials supplied to CSPI's ten consultants 
and on which, presumably, the consultants had based their comments. 
CSPI responded by submitting copies of materials that it characterized 
as ``a standard data set,'' consisting of ten complete documents and 
selected portions of several other documents (19 items altogether) 
drawn from the administrative record for the dry uses final rule.\23\ 
Based on the ``standard data set'' submitted by CSPI, it appears that 
the ten consultants were not provided, however, with all of the Hoechst 
study reports and other relevant supporting information, nor were they 
provided with all of the FDA review memoranda filed in the 
administrative record for the prior approvals of ACK.\24\ For example, 
neither the results of the ACK genetic toxicity testing nor FDA's final 
pathology review memorandum (Ref. 21), which articulated FDA's 
resolution of the outstanding questions regarding missing data and 
incomplete initial reporting of histopathology results raised in 
earlier FDA review memoranda, were included in CSPI's ``standard data 
set.''
---------------------------------------------------------------------------

    \23\ The ten complete documents in CSPI's ``standard data set'' 
were six FDA review memoranda, including the final review memorandum 
from FDA's Cancer Assessment Committee; the dry uses final rule (53 
FR 28379); FDA's response to CSPI's objections to the dry uses final 
rule (57 FR 6667); and two letters addressed to Hoechst from an 
independent pathology lab, supplying additional information 
regarding histopathology data (one letter in regard to a long-term 
study in rats, the other in regard to a long-term study in mice). 
The other items in CSPI's ``standard data set'' consisted primarily 
of narrative sections from, or excerpts from various tables (e.g., 
mortality data, tumor incidence data) included in, the study reports 
for the three long-term feeding studies of ACK in rodents.
    \24\ Judging from their remarks, some of CSPI's ten consultants 
may have been under the impression that all of the data and 
information on ACK had been made available to them. For example, one 
of these individuals stated: ``I agree strongly with [CSPI's] 
evaluation that the available data on this compound is at best 
incomplete * * * I could not find any information related to 
mutagenicity or other genotoxicity or any studies on reproduction 
and development.'' Another of CSPI's consultants also made similar 
remarks regarding the apparent lack of ACK genetic toxicity data.
    However, as noted previously in this document, the ACK toxicity 
data base submitted to the original petition for ACK included the 
results of six genetic toxicity tests and four studies of 
reproductive or developmental toxicity. The agency concluded that 
the results of the genetic toxicity tests did not indicate ACK-
induced genotoxic effects and that the results of the reproduction 
and teratology studies produced no evidence of ACK-related 
teratogenic or adverse reproductive effects (see 53 FR 28379 at 
28380).
---------------------------------------------------------------------------

    As previously noted, most of the letters from CSPI's ten 
consultants did not raise specific issues regarding either the long-
term testing of ACK or other safety data relevant to FDA's evaluation 
of the present petition; only one consultant provided detailed 
criticism of FDA's interpretation of the data. FDA's analysis of the 
few specific points raised in letters from the ten consultants is 
discussed below, along with FDA's analysis of the issues raised in 
CSPI's NTP nomination package.
5. Hoechst's Submission
    In response to the letters from CSPI's ten consultants, Hoechst 
transmitted to FDA copies of letters from two CSPI

[[Page 36357]]

consultants to whom the firm had provided supplementary information 
regarding the toxicity testing of ACK. In their letters, these two 
individuals stated that, after reviewing additional information 
provided to them by Hoechst, they had concluded that the long-term 
testing of ACK was adequate and that the test results did not indicate 
that ACK was a carcinogen.
    Hoechst also submitted to FDA copies of the materials it had 
provided to the two CSPI consultants for review. These materials 
included several documents from the administrative record for the dry 
uses final rule as well as a copy of the dry uses final rule. Also 
included in Hoechst's information package was a copy of a document 
entitled ``Executive Summary,'' a document that, according to Hoechst, 
was a summary of toxicology information on ACK that had been submitted 
to Health Canada as part of a petition for the use of ACK; and a book, 
entitled Acesulfame Potassium.\25\
---------------------------------------------------------------------------

    \25\ This book, co-edited by a Hoechst scientist and a professor 
at a German university, discusses various studies of ACK submitted 
in the original petition, including genetic toxicity studies, acute 
studies, the three long-term feeding studies in rodents referred to 
previously in this document, a subchronic feeding study, 
reproduction and teratology studies, metabolism studies and others. 
The book also discusses several additional studies of ACK (e.g., 
additional genetic toxicity studies), conducted after FDA's initial 
approval decision, that were submitted to the present petition and 
have been discussed previously in this document.
---------------------------------------------------------------------------

    Because the additional letters from these two particular 
consultants provided no data or other substantive information, FDA 
regards them solely as further elaboration of the earlier remarks from 
the two individuals in question. No further discussion of any of these 
remarks is necessary.
6. Other Submissions
    FDA also received several letters from trade groups and other 
organizations urging FDA to approve the present petition. Because none 
of these letters provided any substantive information, no further 
discussion of these submissions is necessary.

B. Analysis of Specific Issues Raised in the Comments

1. AAA Test Results
    CSPI, in its fourth submission, and two of CSPI's ten consultants, 
commented on the results of short-term toxicity tests of ACK's 
breakdown product, AAA, and raised the issue of AAA's possible 
carcinogenic potential.\26\ FDA agrees that the results of the short-
term studies of AAA raised concerns that required resolution. As 
discussed previously, the agency carefully evaluated the data from the 
short-term toxicity tests of AAA, along with other data and information 
from the petition and in its files. As discussed previously, FDA has 
concluded that AAA is highly unlikely to pose a significant cancer risk 
to individuals consuming ACK-sweetened beverages; none of the 
information in the comments provides a basis to reconsider that 
conclusion. Because the agency's detailed analysis of the issue of 
AAA's possible carcinogenic potential has already been presented (see 
sections III.D.2.b.v and vi of this document), that analysis will not 
be repeated here. The agency's analysis of the remaining issues raised 
in the comments on the present petition follows.
---------------------------------------------------------------------------

    \26\ One of these individuals referred to AAA as a ``metabolic 
breakdown product.'' FDA notes, however, that AAA has not been shown 
to be a metabolite of ACK. As discussed previously in this document, 
the ACK toxicity data base submitted to the original petition for 
ACK included the results of 11 metabolism studies. FDA carefully 
evaluated the results of these studies and concluded that they 
revealed no evidence that ACK was metabolized (53 FR 28379 at 28380, 
see also Ref. 4).
---------------------------------------------------------------------------

2. ACK Test Results
    In its NTP nomination package, CSPI again raised some of the same 
questions regarding the adequacy of, and the results from, the long-
term testing of ACK that it raised in its previous objections to the 
dry uses final rule; CSPI also raised some new points with respect to 
the safety testing of ACK. CSPI's NTP nomination package is clearly 
addressed to NTP and is not written as a comment, per se, on the 
present petition; the narrative in CSPI's NTP nomination package 
focuses on the differences between the designs of, and procedures used 
in, the long-term feeding studies of ACK and specific elements of NTP 
study designs or other ``NTP standards.'' Nevertheless, FDA has assumed 
that CSPI's NTP nomination package constitutes the ``detailed analysis 
of the safety data on ACK'' that CSPI had intended to send to the 
agency at a future date and that FDA had indicated it would treat as a 
comment on the present petition. Thus, FDA has attempted to extract 
from CSPI's NTP nomination package those remarks on specific issues 
that could be construed as comments on the present petition.
    As noted previously, there is considerable overlap between the 
specific issues raised by certain of CSPI's ten consultants and those 
raised by CSPI. Because CSPI's NTP nomination package provides the most 
detailed discussion of specific issues, those remarks will be the focus 
of FDA's response. Where the other parties have raised additional 
points or points that differ substantively from those raised by CSPI, 
FDA will indicate that in its discussion.
    a. The second rat study. In its original evaluation of the safety 
of ACK, FDA reviewed a long-term study conducted in CPB-WU Wistar rats 
in which ACK was administered at 0, 0.3, 1.0, or 3.0 percent in the 
test diet (the ``second rat study''). In the preamble to the dry uses 
final rule, the agency concluded that this study was adequate for an 
evaluation of a food additive and that it demonstrated the safety of 
acesulfame potassium (see 53 FR 28379 at 28380). Implicit in FDA's 
determination of the adequacy of the second rat study was that the 
dosing levels in this study were appropriate (see 57 FR 6667 at 6669).
    i. Issues raised previously--(1) Appropriateness of the dosing. 
CSPI's NTP nomination package asserts that the second rat study was 
inadequate because the highest dose tested (3 percent in the diet) was 
too low. To support its assertion, CSPI compares the dosing regimen 
used in the second rat study with NTP ``requirements'': ``NTP requires 
that long-term feeding studies be carried out at the minimally toxic 
dose (MTD), which is functionally equivalent to the maximum tolerated 
dose * * *.'' CSPI also states that ``NTP requires that when a test 
chemical is administered in the diet, the high dose should not exceed 5 
percent of the diet, but use of a 5 percent dose could meet NTP 
standards. Since rats in the subchronic test tolerated 10 percent 
acesulfame potassium in the diet with what were reported as only 
minimal effects* * *, 5 percent should have been the highest dose 
tested in the two rat studies.''\27\ CSPI's submission does not, 
however, contain or identify any data or other evidence to establish 
that the dosing used in the second rat study was, in fact, too low to 
permit an assessment of ACK's carcinogenic potential.
---------------------------------------------------------------------------

    \27\ FDA notes that, in the subchronic study, ACK was 
administered at dose levels of 0, 1.0, 3.0, or 10.0 percent in the 
diet. ACK-related reductions in body weight of greater than 10 
percent, along with various other effects, were observed in the 10 
percent dose group. Body weight reductions were also observed in the 
3 percent dose group, but such reductions were less than 10 percent. 
Based on the findings in the 10 percent and 3 percent dose groups, 
Hoechst chose to use 3 percent as the highest dose level in the 
long-term study; there are no data to suggest that 5 percent was 
required.
---------------------------------------------------------------------------

    CSPI implies that, in order for long-term toxicity 
(carcinogenicity) testing to be valid, it must conform to NTP 
``requirements.'' FDA does not agree. The NTP document cited by 
CSPI\28\

[[Page 36358]]

establishes standardized protocol elements and reporting formats for 
certain toxicity and carcinogenicity tests conducted by contract 
laboratories under the auspices of the NTP program. The NTP document 
does not establish criteria for evaluating the scientific validity of 
toxicity and carcinogenicity tests in general, nor does it establish 
regulatory requirements with respect to safety decisions on food 
additives. The NTP document provides specifications that must be met in 
order for the results of a particular toxicity study to be included in 
the NCI/NTP data base (described previously in this document).
---------------------------------------------------------------------------

    \28\ This document is entitled ``Specifications for the Conduct 
of Studies to Evaluate the Toxic and Carcinogenic Potential of 
Chemical, Biological and Physical Agents in Laboratory Animals for 
the National Toxicology Program (NTP).''
---------------------------------------------------------------------------

    FDA notes that the agency's own guidelines, ``Toxicological 
Principles for the Safety Assessment of Direct Food Additives and Color 
Additives Used in Food'' (the FDA Redbook), do not establish regulatory 
requirements or requirements for establishing the scientific validity 
of testing. Rather, the Redbook represents the agency's best advice to 
manufacturers of food and color additives on how to satisfy the legal 
safety standard of ``reasonable certainty * * * that a substance is not 
harmful'' (see Sec. 170.3(i)); and contains general toxicological 
principles that are to be applied using good scientific judgment.
    It is important to note that although the details provided in the 
NTP document differ from those provided in the Redbook, a study that 
follows either the NTP ``specifications'' or the Redbook guidance\29\ 
and is conducted in accordance with good laboratory practices will 
generally be appropriate for use in a safety evaluation. Strict 
adherence to any particular set of guidelines is not necessary, 
however, to ensure either scientific validity or suitability for a 
regulatory safety decision. Accordingly, in reaching a final decision 
on the safety of a food additive, FDA considers all of the relevant 
data and information available, including the design of, and results 
from, toxicity testing. The suitability and validity of any particular 
toxicity study submitted in support of a food additive is evaluated on 
its own merits, using good scientific judgment, by FDA.
---------------------------------------------------------------------------

    \29\ Other guidelines, such as those issued by EPA or the 
Organization for Economic Cooperation and Development (OECD), are 
also frequently used as resources in the design, conduct, and 
evaluation of toxicological tests (see for example, Ref. 22).
---------------------------------------------------------------------------

    The agency notes that, in its objections to the dry uses final 
rule, CSPI raised the same issue regarding the adequacy of the dosing 
in the second rat study, and FDA addressed this issue in its response 
to CSPI's objections (57 FR 6667 at 6668 and 6669). The agency 
incorporates that discussion, in full, into the safety determination on 
the present petition. Because CSPI has presented no new evidence to 
support its opinion regarding the adequacy of the dosing in this study, 
nor identified evidence that the agency overlooked in its previous 
evaluations, FDA reaffirms its earlier determination that the dosing in 
the second rat study was adequate for an assessment of the carcinogenic 
potential of acesulfame potassium (57 FR 6667 at 6669, see also 53 FR 
28379, 28380).
    With respect to dosing, one of CSPI's consultants asserted that the 
dose range in the second rat study was too narrow, citing ``[the] 
increased tumorigenesis at even the `lowest' dose used * * *.'' FDA has 
previously concluded, however, that the data from the second rat study 
do not establish an association between tumors and treatment with ACK 
(53 FR 28379 at 28380 and 28381). The issue of tumor incidence in the 
second rat study is also discussed later in this document.
     CSPI, in its NTP nomination package, also implies that the second 
rat study is inadequate because the subchronic testing of ACK, used as 
an aid in determining doses for the second rat study, did not conform 
in each and every respect to the standardized elements in the NTP 
guidelines. Specifically, CSPI stated that a subchronic study was not 
conducted in the same strain of rat as that used in the second rat 
study; CSPI also disagrees with the use, in the subchronic study, of 
fewer dose groups than the number NTP ``requires.''\30\
---------------------------------------------------------------------------

    \30\ CSPI specifically noted that the NTP document stipulates 
the use of five dose groups in addition to controls. FDA notes that 
the use of five dose groups is not a requirement, either for the 
scientific validity of the test, or for utility of the test in 
reaching a regulatory decision. FDA's own Redbook recommends (but 
does not require) the use of at least three dose groups in addition 
to controls; EPA's guidelines for subchronic toxicity testing 
contain a similar recommendation.
---------------------------------------------------------------------------

    FDA disagrees. First, the agency notes that the purposes of 
subchronic testing are generally acknowledged to be twofold: To 
identify likely target organs in longer-term studies and to aid in 
determining doses for the longer-term testing. Second, as previously 
noted, the NTP document does not establish scientific or regulatory 
requirements for either subchronic or long-term toxicity testing, 
including carcinogenicity testing. In particular, the NTP document does 
not establish a subchronic testing regimen that must be followed in 
order for long-term testing to be valid. Moreover, FDA is not aware of 
any relevant guideline, including the NTP document, that states that 
deviations from the guidelines for a subchronic toxicity study 
conducted to determine appropriate dose levels in a subsequent 
carcinogenicity study necessarily invalidates the results of the 
carcinogenicity study.
    Because CSPI has not provided any substantive information to 
support its assertions regarding the effect of the design of the ACK 
subchronic study on the validity of the long-term testing of ACK, it 
has provided no basis for FDA to reconsider its conclusions regarding 
the second rat study. Thus, FDA reaffirms its earlier conclusions that 
the dosing in the second rat study was appropriate for an assessment of 
the carcinogenic potential of ACK and that the study was suitable for a 
safety assessment of ACK (57 FR 6667 at 6669, see also 53 FR 28379 at 
28380).
    (2) Incidence of mammary tumors. In its NTP nomination package, 
CSPI stated that there was an increased incidence of mammary tumors in 
treated females in the second rat study. CSPI also claimed that ``* * * 
FDA discounted these data because [the] incidence was not strongly 
dose-related.'' CSPI thus implies that the lack of a strong dose-
response was the only reason FDA concluded, in its previous evaluation, 
that the incidence of mammary tumors in female rats in the second rat 
study was not ACK-related. CSPI also criticizes the agency's use of 
historical control data in evaluating the results of the second rat 
study and asserts that more information on ``animals or test 
conditions'' (e.g., diets, animal husbandry) should have been obtained 
by FDA before using the data from ``previous studies'' conducted at the 
testing laboratory where the long-term studies of ACK were 
conducted.\31\
---------------------------------------------------------------------------

    \31\ One of CSPI's consultants criticized the petitioner's use 
of historical control data, commenting that the ``historical 
database'' is ``actually very small.'' CSPI's consultant did not, 
however, provide any information to indicate that FDA made 
inappropriate use of the relevant historical control data. (As 
previously noted, FDA's final pathology review memorandum, which 
discusses the agency's use of the historical control data, was 
apparently not included in the materials supplied by CSPI to its ten 
consultants.)
---------------------------------------------------------------------------

    The agency notes that CSPI has previously raised these particular 
points in its objections to the dry uses final rule, and that FDA has 
previously addressed these points at length in responding to CSPI's 
objections (57 FR 6667 at 6674 and 6675). Specifically, in the original 
safety evaluation of ACK, FDA gave careful and detailed consideration 
to the incidence of mammary gland tumors in female rats in the second 
rat study. After a review of

[[Page 36359]]

all the data, the agency concluded that mammary gland neoplasms were 
not associated with treatment with ACK. The preamble to the dry uses 
final rule cited several reasons for this conclusion, including the 
lack of a dose response. However, the agency also took into account the 
lack of evidence of progressive stages of mammary gland neoplasms and 
certain information obtained from historical control data (53 FR 28379 
at 28381, see also Ref. 21).
    With respect to the use of historical control data, the agency 
notes that, as in its objections to the dry uses final rule, CSPI 
mischaracterizes the information on historical controls and fails to 
acknowledge the detailed information on this point that FDA has 
evaluated. In its response to CSPI's objections, the agency noted that 
the historical control data were from the same type of studies 
conducted in the same laboratory, with the same strain of rat, under 
similar conditions, with continuity of pathological standards, and, 
furthermore, were from the same time period as the long-term studies 
evaluated in FDA's original review (57 FR 6667 at 6672 and Ref. 8 of 
that document). CSPI has presented no new information to support its 
allegation that FDA made inappropriate use of the relevant historical 
control data.
    In summary, CSPI has presented no new evidence that would change 
the agency's previous conclusion that the occurrence of mammary gland 
neoplasms was not associated with treatment with ACK, and FDA 
incorporates its earlier discussion of the results of the second rat 
study, in full, into the safety determination on the present petition. 
Because CSPI has presented no new evidence to support its opinion nor 
identified evidence that the agency overlooked in its previous 
evaluations, FDA reaffirms its earlier determination that the data from 
the second rat study do not establish an association between the 
occurrence of neoplasms and treatment with ACK (53 FR 28379 at 28380 
and 28381).
    ii. Issues not raised previously--(1) Incidence of respiratory 
disease. In its NTP nomination package, CSPI claims that the incidence 
of respiratory disease in the animals used in the second rat study was 
too high\32\ and questioned whether this study or the other long-term 
studies of ACK in rodents were adequate: ``The poor health of the 
animals used in the Hoechst studies raises the question as to whether 
any of the test results in the subchronic and chronic studies were good 
enough to be used.'' However, CSPI's submission neither identifies nor 
contains any data or other evidence that establish that the second rat 
study was, in fact, rendered inadequate for an assessment of ACK's 
carcinogenic potential by the incidence of respiratory disease in the 
test animals.
---------------------------------------------------------------------------

    \32\ CSPI presents some figures for the incidence of pneumonia 
in the rats in the second study that are apparently derived from 
information in the final report for this study, a document not 
included in CSPI's NTP nomination package.
---------------------------------------------------------------------------

    In its original evaluation of the safety of ACK, FDA carefully 
considered all of the data and information relevant to an evaluation of 
the long-term testing of ACK, including the general health of, and the 
incidence of respiratory disease in, test animals. In the case of the 
second rat study, FDA determined that the mortality rate was low in all 
dose groups and the signs of chronic respiratory disease randomly 
distributed (Refs. 21 and 23). Only in the case of the first rat study 
did FDA conclude that the incidence of respiratory disease in test 
animals confounded the test results to such an extent that such 
incidence contributed to a finding that the study was inadequate for 
assessing the safety of ACK (53 FR 28379 at 28380, see also Ref. 24). 
Because CSPI has not presented any new evidence to support its 
allegation nor has the organization identified evidence that the agency 
overlooked in its previous evaluations, FDA reaffirms its earlier 
determination that the second rat study was adequate for an assessment 
of the carcinogenic potential of acesulfame potassium.
    (2) Assignment of animals to test groups. CSPI's NTP nomination 
package also raises a question regarding the procedure used to assign 
animals to the various test groups in the second rat study. CSPI 
implies that improper assignment procedures were used, which confounded 
the results of the second rat study. CSPI does not, however, provide 
any data or other information to support its speculation.\33\
---------------------------------------------------------------------------

    \33\ In its NTP nomination package, CSPI remarks: ``* * * the 
likelihood that animals were of different ages when exposure to the 
test agent began, and that female animals may have been considerably 
older than males, makes it difficult to know what to make of the 
data.'' While CSPI speculates, at length, on the ages of the animals 
in the subchronic study, CSPI does not provide any substantive 
information to support its claims regarding the long-term study, nor 
does the organization provide an explanation of the significance of 
its allegations.
---------------------------------------------------------------------------

    In its original evaluation of the safety of ACK, FDA carefully 
considered all of the data and information relevant to an evaluation of 
the long-term testing of ACK, including the question of whether the 
assignment procedures or other aspects of the study designs compromised 
the suitability of the studies for an assessment of ACK's carcinogenic 
potential (Ref. 23). FDA concluded that the second rat study was 
adequate for an assessment of ACK's carcinogenic potential (Ref. 24, 
see also 53 FR 28379, 28380, and 57 FR 6667 at 6669). Because CSPI, in 
support of its allegations, has neither presented evidence that has not 
already been evaluated by the agency nor identified evidence that the 
agency overlooked in its previous evaluations, FDA reaffirms its 
earlier conclusion that the second rat study was adequate for an 
assessment of ACK's carcinogenic potential.
    b. The mouse study. In concluding that ACK had been shown to be 
safe, FDA reviewed a long-term study conducted in Swiss mice in which 
ACK was administered at 0, 0.3, 1.0, or 3.0 percent in the test diet 
(``the mouse study''). FDA concluded that the results of this study 
showed no association between neoplastic disease and treatment with ACK 
(53 FR 28379 at 28380). In the preamble to the dry uses final rule, the 
agency explicitly discussed the adequacy of the mouse study with 
respect to study duration. FDA concluded that the length of the study 
was adequate because it had been conducted for the majority of the 
animals' lifespan (53 FR 28379 at 28380; see also 57 FR 6669 at 6670). 
Implicit in FDA's determination of the mouse study's adequacy was that 
the dosing levels in this study were appropriate (57 FR 6667).
    i. Issues raised previously--(1) Adequacy of the study length. In 
its NTP nomination package, CSPI asserts that the mouse study was 
inadequate because the study was too short. To support its assertion, 
CSPI again refers to NTP ``requirements'': ``NTP generally requires 
that long-term studies on rats and mice be carried out for a 104-week 
period. Hoechst's study in mice lasted only 80 weeks.'' CSPI also 
presents some figures for survival levels in the various test groups 
(apparently derived from information in the final report for the mouse 
study, a document not included in CSPI's NTP nomination package) and 
remarks that ``survival of the mice was very high at 80 weeks.'' CSPI 
implies that the survival statistics suggest that the study was not 
conducted for the majority of the animals' lifespan. However, CSPI 
provides no data or other evidence to support its view.
    FDA disagrees with CSPI's comments regarding the length of the 
mouse study. First, as previously noted in this document, the NTP 
document cited by CSPI does not establish either scientific or 
regulatory requirements. Second, in its original evaluation of the 
safety of ACK, FDA carefully considered all of the data and information 
relevant to an

[[Page 36360]]

evaluation of the long-term testing of ACK, including the duration of, 
and survival data from, the mouse study. As previously noted, FDA 
concluded that length of the study was adequate because it had been 
conducted for the majority of the animals' lifespan (see 53 FR 28379 at 
28380, see also Ref. 24.) Specifically, the agency found that at the 
time the study was conducted, survival of the Swiss strain of mice 
tended to decline severely between 18 and 24 months of age; thus, at 
that time, 80 weeks was representative of a time period corresponding 
to the majority of the animals' lifespan (Ref. 24).
    CSPI previously raised this issue in its objections to the dry uses 
final rule, and the agency previously discussed this issue in 
responding to CSPI's objections (57 FR 6667). FDA incorporates that 
discussion, in full, into the safety determination on the present 
petition. Because CSPI has not identified any evidence that the agency 
overlooked in its previous evaluations, FDA reaffirms its earlier 
determination that the mouse study was of adequate duration for an 
assessment of the carcinogenic potential of ACK.
    (2) Appropriateness of dosing. CSPI, in its NTP nomination package, 
comments on the appropriateness of the dosing in the mouse study: ``* * 
* the high survival at 80 weeks of mice fed 3% acesulfame potassium in 
the diet suggests that a higher dose might have been more in keeping 
with NTP recommendations.'' CSPI provides no other further explanation 
of the significance of its remarks, nor does it provide any data or 
other information that would establish that the dosing in the mouse 
study was too low to permit an assessment of ACK's carcinogenic 
potential. CSPI previously questioned the adequacy of the dosing in the 
mouse study in its objections to the dry uses final rule, and the 
agency previously discussed this issue in responding to CSPI's 
objections (57 FR 6667). FDA incorporates that discussion, in full, 
into the safety determination on the present petition. Because CSPI has 
presented no new evidence to support its opinion nor identified 
evidence that FDA overlooked in its previous evaluations, FDA reaffirms 
its earlier determination that the dosing in the mouse study was 
appropriate for an assessment of the carcinogenic potential of 
acesulfame potassium (see 57 FR 6667 at 6669).
    ii. Issues not raised previously--(1) Incidence of respiratory 
disease. In its NTP nomination package, CSPI notes that respiratory 
infections occurred in the mice, but offers no specific supporting 
information.\34\ In particular, CSPI neither identifies nor provides 
any data or other evidence regarding the actual incidence of 
respiratory infections in the mice, nor does it provide any information 
that would establish that the mouse study was rendered inadequate for 
an assessment of ACK's carcinogenic potential by the alleged incidence 
of respiratory disease in the test animals.
---------------------------------------------------------------------------

    \34\ As noted previously in this document, CSPI questions, in 
its NTP nomination package, the health of the test animals in all of 
the long-term studies of ACK in rodents. However, CSPI also cites 
the high survival rates of the test animals in the mouse study in 
support of some of the organization's criticisms of this study. The 
agency notes that CSPI's positions regarding animal health and 
survival rates in the mouse study are not entirely consistent.
---------------------------------------------------------------------------

    FDA notes that, in its original evaluation of the safety of ACK, 
the agency carefully considered all of the data and information 
relevant to an evaluation of the long-term testing of ACK, including 
the health of the test animals (Ref. 23). CSPI has presented no 
evidence to support its claim that has not already been evaluated by 
the agency nor identified evidence that the agency overlooked in its 
previous evaluations. Thus, FDA reaffirms its earlier conclusion that 
the mouse study was suitable for an assessment of ACK's carcinogenic 
potential (see 53 FR 28379 at 28380, and 57 FR 6667 at 6669).
    (2) Histopathology data. CSPI also criticizes aspects of the 
histopathological examinations in the mouse study. CSPI specifically 
compares the extent of the histopathology review of tissues from 
animals from the low and mid-dose test groups with ``NTP 
requirements.'' CSPI implies that the histopathology review was not 
extensive enough and, thus, obscured the results of the mouse study. 
CSPI does not, however, provide any data or other information that 
would establish that the histopathological examinations of tissues from 
the animals in the mouse study were inadequate for an assessment of 
ACK's carcinogenic potential.
    FDA notes that, in its original evaluation of the safety of ACK, 
the agency carefully considered all of the data and information 
relevant to an evaluation of the long-term testing of ACK, including 
the histopathology data from the mouse study. FDA concluded both that 
the mouse study was adequate for an assessment of ACK's carcinogenic 
potential and that the results of the study showed no association 
between neoplastic disease and treatment with ACK (53 FR 28379 at 28380 
and 57 FR 6667 at 6669, see also Ref. 24). Again, because CSPI has 
presented no evidence to support its assertions that has not already 
been evaluated by the agency nor has CSPI identified evidence that the 
agency overlooked in its previous evaluations, FDA reaffirms its prior 
conclusion that the mouse study was suitable for an assessment of ACK's 
carcinogenic potential.
    (3) Time-to-tumor. In its NTP nomination package, CSPI also claims 
that the data in the mouse study showed that ACK caused tumors: ``[i]n 
the mouse study, there was an early time-to-tumor reported for first 
tumors in treated animals relative to first tumors in controls.'' 
However, CSPI provides no additional data or other information to 
support this claim, nor does it provide further explanation of the 
significance of this alleged time-to-tumor differential.
    In the original safety evaluation of ACK, FDA carefully considered 
all of the data in the mouse study, including data in the study report 
that showed an apparent ACK-related decreased time-to-tumor for first 
tumors. After an interim review of all the data, the agency concluded 
that the only finding of possible significance was an increase in 
lymphocytic leukemia in female mice in the highest dose group (Ref. 
25). After detailed consideration of this reported finding, FDA 
concluded that this finding was not treatment-related and that no 
increase in neoplastic disease of the lymphoreticular system could be 
attributed to ACK (Ref. 24).
    Because CSPI has presented no new evidence to support its opinion 
nor identified evidence that the agency overlooked in its previous 
evaluations, it has provided no basis for FDA to change its previous 
conclusions regarding the results of the mouse study. Thus, FDA 
reaffirms its earlier determination that the data from the mouse study 
do not establish an association between neoplasia and treatment with 
ACK (see 53 FR 28379 at 28380 and 57 FR 6667 at 6669).
    c. The first rat study. In its evaluation of the original petition 
for the use of ACK, the agency reviewed a long-term study conducted in 
CIVO-bred Wistar rats in which ACK was administered at 0, 0.3, 1.0, or 
3.0 percent in the diet (the ``first rat study''). In the preamble to 
the dry uses final rule, the agency concluded that the data from this 
study did not establish a carcinogenic effect of ACK (53 FR 28379 at 
28380). However, the agency further concluded, because of deficiencies 
and confounding factors in this study (e.g., a high incidence of 
respiratory disease in the test animals), that it was ``inadequate for 
assessing the carcinogenic potential of the test compound or for any 
other purposes of

[[Page 36361]]

a safety evaluation'' (53 FR 28379 at 28381).
    Issues raised previously. In its NTP nomination package, CSPI 
asserts that, despite the prevalence of chronic respiratory disease in 
the test animals in the first rat study, the test results were 
suggestive of a carcinogenic effect of ACK.\35\ Specifically, CSPI 
claims that the data in the first rat study showed a dose-dependent 
effect on incidence of lymphoreticular cancers of pulmonary origin and 
on time-to-tumor. In support of its claims, CSPI cites a single FDA 
interim review memorandum (Ref. 23). CSPI also asserts that the agency 
made inappropriate use of historical control data in evaluating the 
results of the first rat study.\36\ With respect to the use of 
historical control data, CSPI merely expresses its opinion that more 
information on ``animals and test conditions'' (e.g., diets and animal 
husbandry) should have been obtained by FDA before using the data from 
``previous studies'' conducted at the testing laboratory where the 
long-term studies of ACK were conducted.
---------------------------------------------------------------------------

    \35\ Several of CSPI's ten consultants made similar remarks. 
None of these individuals, however, provided any substantive 
information in support of their remarks.
    \36\ Importantly, as in its objections to the dry uses final 
rule, CSPI mischaracterizes the information on historical controls 
and fails to acknowledge the information on this point that FDA 
evaluated. The agency has previously discussed, in detail, its use 
of historical control data in the evaluation of the first rat study 
in responding to CSPI's objections to the dry uses final rule. In 
its response to CSPI's objections, the agency noted that the 
historical control data were from the same type of studies conducted 
in the same laboratory, with the same strain of rat, under similar 
conditions, with continuity of pathological standards, and, 
furthermore, were from the same time period as the first rat study 
(57 FR 6667 at 6672).
---------------------------------------------------------------------------

    The agency notes that the issue of a possible dose-dependent effect 
of ACK on the incidence of lymphoreticular tumors and on time-to-tumor 
was raised by CSPI in its letter to FDA dated September 23, 1987, and 
this issue was addressed by the agency in the preamble to the dry uses 
final rule (53 FR 28379). Specifically, the agency noted that, in the 
first rat study, there was a slightly higher incidence, and earlier 
appearance, of lymphoreticular tumors in dosed rats than in the 
concurrent control group. However, the agency concluded that under the 
circumstances of severe chronic respiratory disease, sampling 
limitations, and the very high rate of spontaneously-occurring lung 
tumors in this strain of rat, no conclusions could be made regarding 
any effect of ACK on the lungs (53 FR 28379 at 28380; see also Ref. 
24).\37\ FDA also notes that CSPI previously raised this particular 
issue in its objections to FDA's original approval decision on ACK, and 
the agency discussed these issues, at length, in responding to CSPI's 
objections (57 FR 6667 at 6671 and 6672). FDA incorporates those 
discussions, in full, into the safety determination on the present 
petition. Because CSPI has presented no new evidence to support its 
opinion nor identified evidence that the agency overlooked in its 
previous evaluations that would change the outcome of those 
evaluations, FDA reaffirms its earlier determination that the data from 
the first rat study do not establish a carcinogenic effect of ACK.
---------------------------------------------------------------------------

    \37\ Because the first rat study was inadequate for use in 
assessing the carcinogenic potential of ACK, the petitioner 
conducted a second long-term study in a different strain of rat. 
This second rat study did not show lymphoreticular tumors in the 
lungs (53 FR 28379 at 28380).
---------------------------------------------------------------------------

C. Summary of FDA's Response to Comments

    In determining that ACK is safe for use in nonalcoholic beverages, 
FDA carefully considered all of the data and information in the present 
petition, as well as other information in its files, including relevant 
information from previous petitions for ACK. FDA has also carefully 
considered all of the issues raised in the comments on the present 
petition.
    As previously noted in this document, many of the specific issues 
raised in the comments on the present petition are the same as those 
raised in earlier objections to the dry uses final rule, and the agency 
has previously considered and responded to these issues in detail (see 
57 FR 6667). Also as noted, the comments supply no new information that 
would change any of the agency's prior conclusions on any of the issues 
previously raised. Likewise, with respect to specific issues raised in 
the comments on the present petition that have not been raised 
previously, the comments neither provide new evidence nor identify 
evidence that FDA has overlooked that would change the agency's 
conclusion that the use of ACK in nonalcoholic beverages is safe.
    Because no outstanding issues in the comments undermine FDA's 
determination of safety, FDA is denying the requests that: (1) The 
petitioner be required to conduct additional testing of ACK or AAA, (2) 
the present petition be denied, and (3) all existing regulations 
permitting the use of ACK in food be revoked.

V. Conclusion of Safety

    FDA has evaluated the data in the petition, published scientific 
literature, and other relevant material from its files and concludes 
that the use of ACK in nonalcoholic beverages is safe. Therefore, the 
agency concludes that Sec. 172.800 should be amended as set forth 
below.
     In accordance with Sec. 171.1(h) (21 CFR 171.1(h)), the petition 
and the documents that FDA considered and relied upon in reaching its 
decision to approve the petition are available for inspection at the 
Center for Food Safety and Applied Nutrition by appointment with the 
information contact person listed above. As provided in Sec. 171.1(h), 
the agency will delete from the documents any materials that are not 
available for public disclosure before making the documents available 
for inspection.

VI. Environmental Impact

    The agency has carefully considered the potential environmental 
effects of this action. FDA has concluded that the action will not have 
a significant impact on the human environment, and that an 
environmental impact statement is not required. The agency's finding of 
no significant impact and the evidence supporting that finding, 
contained in an environmental assessment, may be seen in the Dockets 
Management Branch (address above) between 9 a.m. and 4 p.m., Monday 
through Friday.

VII. Paperwork Reduction Act

    This final rule contains no collections of information. Therefore, 
clearance by the Office of Management and Budget under the Paperwork 
Reduction Act of 1995 is not required.

VIII. Objections

    Any person who will be adversely affected by this regulation may at 
any time on or before August 5, 1998, file with the Dockets Management 
Branch (address above) written objections thereto. Each objection shall 
be separately numbered, and each numbered objection shall specify with 
particularity the provisions of the regulation to which objection is 
made and the grounds for the objection. Each numbered objection on 
which a hearing is requested shall specifically so state. Failure to 
request a hearing for any particular objection shall constitute a 
waiver of the right to a hearing on that objection. Each numbered 
objection for which a hearing is requested shall include a detailed 
description and analysis of the specific factual information intended 
to be presented in support of the objection in the event that a hearing 
is held. Failure to include such a description and analysis for any 
particular objection shall constitute a waiver of the right to a 
hearing on the

[[Page 36362]]

objection. Three copies of all documents shall be submitted and shall 
be identified with the docket number found in brackets in the heading 
of this document. Any objections received in response to the regulation 
may be seen in the Dockets Management Branch between 9 a.m. and 4 p.m., 
Monday through Friday.

IX. References

    The following sources are referred to in this document. References 
marked with an asterisk (*) have been placed on display in the Dockets 
Management Branch (address above) and may be seen by interested persons 
between 9 a.m. and 4 p.m., Monday through Friday. References without an 
asterisk are not on display; they are available as published articles, 
books, and reports.
     *1. Memorandum, from M. DiNovi, Chemistry Review Branch, to P. 
Hansen, Biotechnology Policy Branch, dated April 28, 1994.
    *2. Memorandum to the file FAP 0A4212, from M. DiNovi, K. 
Ekelman, and P. Hansen, dated June 3, 1998.
    *3. Memorandum, from M. DiNovi, Chemistry Review Branch, to P. 
Hansen, Biotechnology Policy Branch, dated November 9, 1994.
    *4. Memorandum, from K. Ekelman, Division of Health Effects 
Evaluation, to P. Hansen, Regulatory Policy Branch, dated June 2, 
1998.
    5. Green, W. L., ``Mechanisms of Action of Antithyroid 
Compounds,'' pp. 77-87 in: The Thyroid, edited by S. C. Werner and 
S. H. Ingbar, Harper & Row, New York, 1978.
    6. Hill, R. N. et al., ``Thyroid Follicular Cell 
Carcinogenesis,'' Fundamental and Applied Toxicology, 12:629-697, 
1989.
    *7. Report, Borzelleca, J. F., C. C. Capen, M. S. Christian, and 
B. N. LaDu, ``Summary and Consensus of the Acesulfame K Scientific 
Expert Panel on the Safety of Acetoacetamide-N-Sulfonic Acid and 
Acetoacetamide,'' dated October 13, 1992.
    *8. Letter, from C.C. Capen, Ohio State University, to J. 
Simplicio, Hoechst-Celanese Corp., dated December 6, 1991.
    9. Gaylor, D. W., and R. L. Kodell, ``Linear Interpolation 
Algorithm for Low Dose Assessment of Toxic Substances,'' Journal of 
Environmental Pathology and Toxicology,  4:305-315, 1980.
    10. National Academy of Sciences/National Research Council, 
``Risk Assessment in the Federal Government: Managing the Process,'' 
Washington, DC, 1983.
    11. Lorentzen, R. J., ``FDA Procedures for Carcinogenic Risk 
Assessment,'' Food Technology, pp. 108-111, 1984.
    12. Gold, L.S. et al., ``Target Organs in Chronic Bioassays of 
533 Chemical Carcinogens,'' Environmental Health Perspectives, 
93:233-246, 1991.
    13. McConnell, E. E., ``Thyroid Follicular Cell Carcinogenesis: 
Results from 343 2-Year Carcinogenicity Studies Conducted by the 
NCI/NTP,'' Regulatory Toxicology and Pharmacology, 16:177-188, 1992.
    14. IRIS (1995), Cincinnati: Office of Health and Environmental 
Assessment, Environmental Criteria and Assessment Office, EPA.
    15. Curran, P. G., and L. J. DeGroot, ``The Effect of Hepatic 
Enzyme-Inducing Drugs on Thyroid Hormones and the Thyroid Gland,'' 
Endocrine Reviews, 12(2):135-150, 1991.
    16. Donaich, I., ``Aetiological Considerations of Thyroid 
Carcinoma,'' vol. 6, pp. 55-72, in: Tumors of the Thyroid Gland, 
edited by D. Smithers, E & S Livingstone, Edinburgh, 1970.
    17. Capen, C. C. and S. L. Martin, ``Mechanisms that Lead to 
Disease in the Endocrine System in Animals,'' Toxicologic Pathology, 
17:234-249, 1989.
    18. Handbook of Carcinogenic Potency and Genotoxicity Databases, 
edited by L. S. Gold and E. Zeiger, CRC Press, Boca Raton, FL, 1997.
    19. Goddard, M. J., D. J. Murdoch, and D. Krewski, ``Temporal 
Aspects of Risk Characterization,'' Inhalation Toxicology, 7:1005-
1018, 1995.
    20. Kodell, R. L., D. W. Gaylor, and J. J. Chen, ``Using Average 
Lifetime Dose Rate for Intermittent Exposures to Carcinogens,'' Risk 
Analysis, 7:339-345, 1987.
    *21. Memorandum, from F. Hines, Diagnostic Pathology Branch, to 
L. Taylor, Additives Evaluation Branch, dated June 6, 1986.
    22. ``Health Effects Test Guidelines,'' U.S. EPA, June, 1996.
    *23. Memorandum, from L. Taylor, Additives Evaluation Branch, to 
P. McLaughlin, Petitions Control Branch, dated November 17, 1982.
    *24. Memorandum, Cancer Assessment Committee (CAC) (covers 
conferences of November 21, 1983, February 21, 1985, December 12, 
1985, and June 17, 1986, and information in Ref. 25 of this 
document).
    *25. Memorandum, from L. Taylor, Additives Evaluation Branch, to 
Cancer Assessment Committee, dated June 19, 1986.

List of Subjects in 21 CFR 172

    Food additives, Reporting and recordkeeping requirements.
    Therefore, under the Federal Food, Drug, and Cosmetic Act and under 
authority delegated to the Commissioner of Food and Drugs, 21 CFR part 
172 is amended as follows:

PART 172--FOOD ADDITIVES PERMITTED FOR DIRECT ADDITION TO FOOD FOR 
HUMAN CONSUMPTION

    1. The authority citation for 21 CFR part 172 continues to read as 
follows:

    Authority: 21 U.S.C. 321, 341, 342, 348, 371, 379e.
    2. Section 172.800 is amended by adding paragraph (c)(13) to read 
as follows:


Sec. 172.800   Acesulfame potassium.

* * * * *
    (c) * * *
    (13) Nonalcoholic beverages, including beverage bases.
* * * * *

    Dated: June 29, 1998.
Michael A. Friedman,
Acting Commissioner of Food and Drugs.
[FR Doc. 98-17700 Filed 6-30-98; 10:34 am]
BILLING CODE 4160-01-F