You are on page 1of 12

Toxicology Letters 149 (2004) 223–234

Toxicology databases and the concept of thresholds


of toxicological concern as used by the JECFA
for the safety evaluation of flavouring agents
A.G. Renwick∗
Clinical Pharmacology Group, Allergy and Inflammatory Sciences Research Division, School of Medicine,
University of Southampton, Biomedical Sciences Building, Bassett Crescent East, Southampton SO16 7PX, UK

Abstract

Since 1996 the FAO/WHO Joint Expert Committee on Food Additives (JECFA) has evaluated the safety of 1259 flavouring sub-
stances, based on a decision tree that incorporates a series of thresholds of toxicological concern. Safety conclusions are based on
the predicted consequences of metabolism and whether the estimated intake is above or below a threshold of toxicological concern
that is relevant to that compound. Compounds are allocated to one of three structural classes, and the intake compared with a thresh-
old of toxicological concern derived using data from chronic and sub-chronic toxicity studies on compounds in the same structural
class. If the substance is predicted to be metabolised to innocuous products there is no safety concern if the intake is below the
threshold, but suitable toxicity data on the compound or structural analogues are required if the intake exceeds the threshold. If the
substance is not predicted to be metabolised to innocuous products, and the intake is below the appropriate threshold, safety evalu-
ation is based on data on the compound or structural analogues. An additional threshold of 1.5 !g per day, derived from doses of in-
vestigated chemicals giving a calculated cancer risk of one in a million, is applied when appropriate toxicity data are not available.
© 2004 Elsevier Ireland Ltd. All rights reserved.

Keywords: Flavouring agent; Safety; Metabolism; Intake; Toxicity

1. Introduction (ii) a level of exposure so low that risk assessment can


be based on the activities of structural analogues,
The term “threshold” can be confusing because it i.e. a threshold of toxicological concern.
has two different uses in toxicology and risk assess-
ment: A search of the WHO website for “JECFA” (Joint
FAO/WHO Expert Committee on Food Additives)
(i) a dose within the dose–response curve below and “threshold” yields a large number of citations,
which an adverse effect would not be produced, but in nearly all cases these refer to the presence
because of homeostatic processes and or absence of a threshold within the dose–response
relationship. In recent years the JECFA has used the
concept of a “threshold of toxicological concern” for
∗ Tel.: +44-23-8059-4261; fax: +44-23-8059-4262. the safety assessment of flavouring agents, and this
E-mail address: agr@soton.ac.uk (A.G. Renwick). will be the basis of this paper.

0378-4274/$ – see front matter © 2004 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.toxlet.2003.12.034
224 A.G. Renwick / Toxicology Letters 149 (2004) 223–234

An extensive toxicological database is required less than one in a million lifetime risk if present in
for compounds present in the food supply in signif- the diet at 0.5 ppb”, and it was a level “2000 times
icant amounts (Renwick, 1999), but it is generally lower than the dietary concentration at which the vast
recognised that the extent of exposure is an important majority of studied compounds are likely to cause
determinant of the amount of data that is considered non-carcinogenic toxic effect” (Federal Register,
necessary to undertake risk assessment (WHO, 1987). 1993). Although the value would give a de minimis
The draft FDA Redbook II (FDA, 1993) proposed risk even if the compound were a genotoxic carcino-
database requirements for three levels of concern, gen, the Threshold of Regulation would not apply
with each concern level based on a combination of to known carcinogens or substances whose chemical
chemical structure and predicted intake. The concept structures provide reason to suspect that they may be
of a threshold of toxicological concern takes this re- carcinogens, because that would contravene US food
lationship to the point where risk assessment can be law. Subsequent refinements to the use of the thresh-
undertaken in absence of toxicity data on the com- old of toxicological concern concept for food contact
pound under evaluation, providing that the intake is materials are discussed in Cheeseman et al. (1999)
sufficiently low. What represents “sufficiently low” and Munro et al. (2003).
depends on the chemical structure of the compound The threshold of toxicological concern principle
and the available toxicity data on structural analogues was developed significantly by the work of Dr. I.C.
that have been subjected to toxicological evaluation. Munro, who analysed available chronic toxicity data
The threshold of toxicological concern represents an and developed a decision tree based on chemical struc-
important and pragmatic approach to risk assessment, ture and intake. The following text analyses the dif-
and can make a significant contribution to optimising ferent thresholds of toxicological concern developed
the use of toxicological resources. The threshold of by Munro, and their use by the JECFA in the risk as-
toxicological concern is used in situations where there sessment of flavouring agents.
are limited or no data on the compound, but the human
exposure is so low that undertaking toxicity studies
is not warranted, because of the costs incurred in the 2. The JECFA decision tree for the safety
use of animals, manpower and laboratory resources. evaluation of flavouring agents
The threshold of toxicological concern does not
provide the surety of risk assessment based on Flavouring agents represent an enormous challenge
compound-specific data because it is based on as- to risk assessors because there are over 3000 sub-
sessing the probability of whether or not adverse stances, with toxicity data available on only a few
effects would occur at the estimated level of exposure. of these. The intakes of flavouring agents are consid-
However, this apparent difference is not “black and ered to be self-limiting because of their taste prop-
white” because there are always uncertainties in risk erties. Flavouring agents have been evaluated by the
assessment, even when there is a comprehensive and JECFA for many years, but it was only with the in-
complete toxicity database. The use of threshold of troduction of a safety evaluation procedure based on
toxicological concern can be thought of as trading-off the Munro analyses and decision tree (JECFA, 1997)
increased uncertainty against very low exposure. that consideration of large number of structurally re-
A threshold of toxicological concern was initially lated compounds became feasible (see Munro et al.,
applied to migrants from food contact materials and 1998). A procedure for use by the JECFA was pro-
was used as the basis of the US Food and Drug Ad- posed by Munro (1996), and presented to the 44th
ministration Threshold of Regulation for indirect food meeting (JECFA, 1995). The essential details can be
additives (Federal Register, 1993, 1995). Initially, found in JECFA Reports after this date, and in the pub-
the threshold of toxicological concern was based on lications of Munro et al. (1998, 1999). The procedure
defining a level of exposure that would give a low involves four principal components:
risk, even if the compound were to be a carcinogen.
A level in food of 0.5 ppb (parts per billion or !g/kg) (i) allocation of the compound to one of three struc-
was selected because “most known carcinogens pose tural classes,
A.G. Renwick / Toxicology Letters 149 (2004) 223–234 225

(ii) consideration of the consequences of metabolism, • Class II substances are simply intermediate. They
(iii) estimation of intake, are less clearly innocuous than those of class I, but
(iv) comparison of the intake with a series of thresh- do not offer the basis either of the positive indication
olds of toxicological concern. of toxicity or of the lack of knowledge characteristic
Since 1997 the procedure has been applied to a total of those in class III.
of 1259 different compounds (Table 1) organised by • Class III substances are those that permit no strong
the Flavour and Extract Manufacturers Association of initial presumptions of safety, or that may even sug-
the USA into groups of structurally related substances. gest significant toxicity. They thus deserve the high-
est priority for investigation. Particularly when per
2.1. Allocation of the compound to one of three capita intake is high or a significant subsection of
structural classes the population has a high intake, the implied hazard
would then require the most extensive evidence for
The procedure is applicable only to the assessment safety-in-use.
of compounds with known chemical structures, be-
The initial steps in the decision tree (Cramer et al.,
cause the concept of structure–activity relationships
1978) are shown in Fig. 1 to illustrate the approach.
is the basis for allocation to one of three classes with
Although the decision tree (Fig. 2) is unsophisticated
different levels of toxicological concern. Cramer et al.
compared with modern computer-based, predictive
(1978) developed a series of 33 questions into a deci-
structure–activity databases, the differences in po-
sion tree that allocates the chemical structure to one
tency were validated by the analyses performed by
of three classes which were defined as follows:
Munro et al. (1996) to develop the threshold of
• Class I substances are those with structures and re- toxicological concern for each structural class (see
lated data suggesting a low order of oral toxicity. later).
If combined with low human exposure, they should
enjoy an extremely low priority for investigation. 2.2. Consideration of metabolism
The criteria for adequate evidence of safety would
also be minimal. Greater exposures would require Predicted metabolism is an important part of the
proportionately higher priority for more exhaustive JECFA procedure and represents the second step in the
study. process (Fig. 3). The way that the different thresholds

Fig. 1. The initial steps in the decision tree of Cramer et al. (1978) used to assign flavouring agents to different structural classes.
226 A.G. Renwick / Toxicology Letters 149 (2004) 223–234

Fig. 2. An overview of Cramer et al. (1978) structural classification scheme-adapted to show where different structural groups are evaluated
for toxic potential.

of toxicological concern are applied in the procedure is whereas if metabolites are not predicted to be innocu-
dependent on the predicted metabolism. Compounds ous they follow the B-side (JECFA, 2000a). JECFA
that are predicted to be metabolised to innocuous prod- (1997) defined “Innocuous products are products
ucts are evaluated using the A-side of the decision tree, that are known or readily predicted to be harmless

Fig. 3. The initial steps in the procedure used by the JECFA for the safety evaluation of flavouring agents.
A.G. Renwick / Toxicology Letters 149 (2004) 223–234 227

to humans at the estimated intakes of the flavouring one weakness of the JECFA evaluations is the lack of
agents”. The prediction of metabolism is difficult and intake estimates for other geographical and cultural
relies on expert judgement of the members and tem- areas.
porary advisors invited to the JECFA meeting. The The per capita × 10 method is simple and readily
definition used by the JECFA relates to the metabolic applicable to the evaluation of flavouring agents; most
fate at the low levels of intake relevant to the intake importantly it has been shown to be appropriately
as a flavouring agent. In consequence detoxication conservative. Intakes estimated by the method were
processes such as glutathione conjugation would generally higher than the mean and 95th percentile
not be saturated, and would rapidly inactivate po- 14-day average intakes found using a detailed dietary
tentially reactive molecules such as aldehydes and analysis based on food consumption and composition
",#-unsaturated carbonyl compounds. (Hall and Ford, 1999). An alternative intake esti-
The reports of the early meetings at which the mate for flavouring agents is the theoretical added
procedure was used contained general text on the maximum daily intake (TAMDI) which is calculated
metabolism of the different functional groups present assuming that 160 g of food and 324 ml of drink con-
in the compounds evaluated at the meeting (JECFA, tain the flavouring substance at the upper use level
1997, 1999, 2000a), but at later meetings metabolism specified by the Council of Europe (Lambe et al.,
was considered separately for each group of sub- 2002). While such an intake might occur rarely on
stances. A large number of flavouring agents are single days, it is not appropriate for comparison with
aldehydes or ",#-unsaturated carbonyl compounds, thresholds of toxicological concern that are derived
which are reactive chemicals, and the procedure was using data from chronic animal studies (see Renwick
considered applicable to these functional groups be- et al., 2003). The TAMDI estimates for a range of
cause of their extensive detoxication by oxidation and flavouring agents were gross overestimates compared
conjugation reactions (JECFA, 2002b). with the per capita × 10 values, and the latter were
supported by the results of a stochastic model that
2.3. Estimation of intake used data from the Dietary and Nutritional Survey
of British Adults combined with information from
There are a variety of methods available for esti- the Irish National Food Ingredients Database (Lambe
mating the intakes of chemical in food (Kroes et al., et al., 2002).
2002). The method used in the JECFA procedure for
flavouring agents is one of the simplest methods, in 2.4. Comparison of the intake with a series of
which the amount of the compound produced annu- threshold of toxicological concern values
ally is divided across the population that may have
consumed foods containing the compound, a so-called The thresholds of toxicological concern for Cramer
per capita estimate. Information on the annual pro- et al. (1978) structural classes were derived by Munro
duction volume for each flavouring agent in Europe et al. (1996) based on an analysis of data from chronic
and the USA is supplied by the flavour manufacturers. toxicity studies on 137, 28 and 448 compounds in
The annual production volumes are divided by 0.6, classes I, II and III respectively. The database com-
on the assumption that only 60% of actual production prised:
may be reported, and allocated to 10% of the relevant
population (equivalent to 32 million in Europe and (i) NOEL (no observed effect level) values for ro-
24 million in the USA; JECFA, 1997). The intake dents and rabbits; NOEL values for dogs and
estimates of flavouring agents for the USA evaluated other species were not included because the stud-
by the JECFA in the 55th Report (JECFA, 2001), and ies contained too few animals to be statistically
subsequently, were corrected by dividing the reported valid,
poundage by 0.8 rather than 0.6 because about 87% of (ii) NOEL values from sub-chronic studies were di-
the poundage data had been reported in recent surveys vided by a three-fold uncertainty factor (WHO,
in the USA. Although the intake estimates relate to ma- 1994) to convert them into equivalent chronic
jor populations that would consume flavoured foods, NOELs and
228 A.G. Renwick / Toxicology Letters 149 (2004) 223–234

(iii) NOEL values derived by the authors were used The way that these threshold values are used in the
except where these were based on physiological JECFA procedure depends on whether the flavouring
rather than toxicological effects. agent is evaluated using the A-side or the B-side.
Using the A-side for substances that are predicted
The NOEL from a chronic toxicity study in rodents to be metabolised to innocuous products (Fig. 4;
is the usual starting point for the determination of an based on JECFA (2000a)) there would not be a safety
acceptable daily intake (ADI) for an approved food concern if the intake of a flavouring agent were below
additive, and assumes that there is a threshold dose the relevant threshold. If the estimated intake ex-
within the dose–response relationship (see Section 1 ceeds the threshold then the next question determines
to avoid confusion). The cumulative distributions whether the compound is endogenous. There would
of NOELs for each Cramer et al. (1978) structural be no safety concern if the compound is an endoge-
class were plotted and a log-normal distribution was nous body constituent. The JECFA recognised that
fitted. The three distributions were significantly dif- many endogenous compounds have important physio-
ferent from each other (Munro et al., 1996). The 5th logical and other functions and that ingestion of such
percentile values for the classes I, II and III NOEL a compound could give rise to adverse effects. There-
distributions were calculated to be 3.0, 0.91 and fore, within the context of the procedure, endogenous
0.15 mg/kg body weight per day. Therefore, there is a substances were defined by the JECFA as “Interme-
95% probability that the NOEL from a chronic animal diary metabolites normally present in human tissues
bioassay on an unstudied compound would be above and fluids, whether free or conjugated; hormones and
the relevant 5th percentile value. The 5th percentile other substances with biochemical or physiological
NOEL values were converted to corresponding human regulatory functions are not included. The estimated
intakes by dividing by the usual 100-fold uncertainty intake of a flavouring agent that is, or is metabolised
factor (WHO, 1987) that is used in the calculation of to, an endogenous substance should be judged not
ADI values for food additives. The intakes were then to give rise to perturbations outside the physiological
multiplied by 60 to scale to the adult human body range” (JECFA, 1997). If the compound is not an en-
weight. These analyses gave thresholds of toxicolog- dogenous substrate, according to this definition, then
ical concern of 1800, 540 and 90 !g per person per data on the compound, or a related substance, can be
day for structural classes I, II and III. used for the safety evaluation.

Fig. 4. The A-side of the procedure used for compounds predicted to be metabolised to innocuous products.
A.G. Renwick / Toxicology Letters 149 (2004) 223–234 229

Fig. 5. The B-side of the procedure used for compounds not predicted to be metabolised to innocuous products.

Using the B-side for substances that are not pre- Potency Database of Gold and colleagues. There
dicted to be metabolised to innocuous products were 477 animal carcinogens in the database when
(Fig. 5; based on JECFA (2000a)) toxicity data on the Threshold of Regulation was proposed (Federal
the compound, or a related substance, would be used Register, 1993) and 709 in the recent reconsideration
for the safety evaluation if the intake of a flavour- by Cheeseman et al. (1999). The threshold of toxico-
ing agent were below the relevant threshold. A more logical concern of 1.5 !g per day was derived Munro
comprehensive risk assessment, using toxicity data on (1990) from the distribution of estimated one in a
the substance or a close structural analogue, would million cancer risks for a subset of the Gold et al.
be necessary if the intake were above the threshold. (1989) database, which comprised the most relevant
There would not be a safety concern if the intake were studies in rodents, i.e. those that contained two or
below the relevant classes I, II or III threshold, and more oral dose levels and a statistically significant
there was a NOEL that gave an adequate margin of dose–response relationship. This analysis showed that
safety. Additional data would be required if there was there was a 63% probability that the estimated cancer
not a suitable NOEL and the intake exceeded a general risk would be one in a million or less for a dietary
threshold of toxicological concern of 1.5 !g per day. concentration of 0.5 ppb (0.5 !g/kg) if all chemicals
There would not be a safety concern if the intake were were carcinogens, but a 96% probability if only one
less than the threshold of toxicological concern of in ten chemicals were carcinogenic. Thus for any
1.5 !g per day. unstudied compound there is a 96% probability that
The threshold of toxicological concern of 1.5 !g the cancer risk would be one in a million or less at
per day was developed from the work of Rulis and a daily intake of 1.5 !g per person per day. This is
colleagues at the FDA (see Cheeseman et al., 1999) on probably a gross over-estimate of the real probability
the Threshold of Regulation for food contact materials because of the selective nature of compounds that
(see Section 1). The FDA threshold of toxicological are subjected to rodent carcinogenicity testing, and
concern was derived from an analysis of the distribu- the highly conservative linear extrapolation method
tion of intakes giving a one in a million cancer risk used to convert the TD50 (the estimated dose giving
using linear low-dose extrapolation of experimental a 50% incidence of tumours in the animal study) to
data in animals for the chemicals in the Carcinogenic a one in a million risk. In addition, the assumption
230
Table 1
Summary of the results of the application of the procedure by the JECFA to the safety evaluation of flavouring agents
Group Total (n) Class I Class II Class III A-side B-side Below No safety Reference
threshold concern
Ethyl esters 15 15 0 0 15 0 11 15 JECFA (1997)
Isoamyl alcohol and related esters 11 11 0 0 11 0 9 11 JECFA (1997)
Allyl esters 21 0 18 3 0 21 20 20a JECFA (1997)
Saturated aliphatic acyclic linear primary alcohols, 38 38 0 0 38 0 27 38 JECFA (1999)
aldehydes and acids
Saturated aliphatic acyclic branched-chain primary 25 22 3 0 25 0 25 25 JECFA (1999)
alcohols, aldehydes and acids
Aliphatic lactones 35 29 0 6 29 6 26A, 2B 31b JECFA (1999)
Esters of aliphatic acyclic primary alcohols with 32 32 0 0 32 0 31 32 JECFA (1999)

A.G. Renwick / Toxicology Letters 149 (2004) 223–234


branched-chain aliphatic acyclic acids
Esters of aliphatic acyclic primary alcohols with 67 66 1 0 67 0 66 66c JECFA (1999)
aliphatic linear saturated carboxylic acids
Esters of branched-chain terpenoid alcohols and 26 26 0 0 26 0 26 26 JECFA (1999)
aliphatic acyclic carboxylic acids
Saturated aliphatic acyclic secondary alcohols, ketones 39 28 11 0 37 2 35A, 2B 39 JECFA (2000a)
and related saturated and unsaturated esters
Linear and branched-chain aliphatic unsaturated 42 42 0 0 41 1 40A, 1B 41d JECFA (2000a)
unconjugated alcohols aldehydes, acids and related
esters
Aliphatic acyclic and alicyclic terpenoid tertiary 23 22 1 0 22 1 19A, 1B 22e JECFA (2000a)
alcohols and structurally related substances
Carvone and structurally related substances 9 6 3 0 9 0 8 9 JECFA (2000a)
Ionones and structurally related substances 21 21 0 0 13 7 13A, 7B 20f JECFA (2000a)
Aliphatic acyclic and alicyclic "-diketones and related 22 0 22 0 22 0 19 22 JECFA (2000a)
"-hydroxyketones
Substances structurally related to menthol 14 9 5 0 14 0 12 14 JECFA (2000a)
Simple aliphatic and aromatic sulphides and thiols 137 97 34 6 0 137 137 137 JECFA (2000b)
Aliphatic linear primary alcohols, aldehydes and 47 47 0 0 47 0 41 47 JECFA (2000b)
carboxylic acids, acetals and esters containing
additional oxygenated functional groups
Cinnamyl alcohol and related flavouring agents 55 50 5 0 55 0 51 55 JECFA (2001)
Furfuryl alcohol and related flavouring agents 15 0 9 6 0 15 15 15 JECFA (2001)
Phenol and phenol derivatives 48 47 0 1 48 0 47 48 JECFA (2001)
Pulegone and related flavouring agents 6 2 4 0 0 6 6 6 JECFA (2001)
Pyrazine derivatives 41 0 32 9 41 0 41 41 JECFA (2002a)
Aromatic substituted secondary alcohols, ketones and 38 28 6 4 36 2 36A, 2B 37e JECFA (2002a)
related esters
Benzyl derivatives 37 37 0 0 36 1 33A, 1B 37 JECFA (2002a)
Hydroxy- and alkoxy-substituted benzyl derivatives 46 36 10 0 46 0 42 46 JECFA (2002a)
Aliphatic acyclic diols, triols and related substances 31 22 9 0 31 0 22 31 JECFA (2002a)
Aliphatic acyclic acetals 10 10 0 0 10 0 10 10 JECFA (2002a)
Alicyclic primary alcohols, aldehydes, acids and related 26 26 0 0 25 1 25A, 1B 26 JECFA (2002b)
esters
Phenethyl alcohol, aldehyde, acid and related acetals 43 39 0 4 43 0 42 43 JECFA (2002b)
and esters
Sulphur-containing heterocyclic compounds 30 0 21 9 3 27 2A, 27B 30 JECFA (2002b)
Sulphur-substituted furan derivatives 33 0 18 15 0 33 33 33 JECFA (2002b)
Alicyclic ketones, secondary alcohols and related esters 25 6 19 0 25 0 25 25 JECFA (2002b)
Aliphatic secondary alcohols, ketones and related esters 39 11 28 0 39 0 39 39 JECFA (2002b)
Alicyclic, alicyclic-fused and aromatic-fused ring 16 4 0 12 10 6 10A, 4B 16 JECFA (in press)
lactones
Aliphatic di- and trienals and related alcohols, acids 26 26 0 0 13 13 13A, 13B 26 JECFA (in press)
and esters
Aliphatic branched-chain unsaturated alcohols, 32 32 0 0 32 0 31 32 JECFA (in press)

A.G. Renwick / Toxicology Letters 149 (2004) 223–234


aldehydes, acids and related esters
Aliphatic and aromatic ethers 29 9 12 8 29 0 26 29 JECFA (in press)
Hydroxypropenylbenzenes 9 6 0 3 9 0 8 9 JECFA (in press)
Total 1259 902 271 86 979 279 1174 1249
a Allyl furoate was deferred because of inadequate data at step B4; a conclusion of no safety concern was reached when step B5 was added to the procedure (JECFA, 1999).
b Four ",#-unsaturated lactones were deferred until a later meeting, at which the application of the procedure to such compounds was considered (JECFA, 2001).
c One ",#-unsaturated ester was deferred until a later meeting, at which the application of the procedure to such compounds was considered (JECFA, 2001).
d The evaluation of one compound on the B-side was deferred pending evaluation of the results of a 90-day study.
e One substance evaluated using the B-side had an intake below its structural class threshold but lacked appropriate toxicity data and the intake was more than 1.5 !g per

day so that additional data were required for an evaluation.


f One of the substances submitted for evaluation was considered not to be a member of the group and was not evaluated.

231
232 A.G. Renwick / Toxicology Letters 149 (2004) 223–234

that only one in ten untested compounds would be although a total of 471 flavouring agents had reported
a carcinogen is probably conservative (Fung et al., intakes below this level.
1995).

4. Conclusions
3. The conclusions reached by the JECFA using
the procedure for the safety evaluations of The JECFA procedure based on the work of Munro
flavouring agents and his colleagues (Munro, 1996; Munro et al., 1996)
provides a practical method for the safety evaluation
The JECFA first used the procedure at a meeting of flavouring agents. The conservative nature of the
in 1996 (JECFA, 1997) to evaluate groups of ethyl, thresholds of toxicological concern compared with the
isoamyl and allyl esters using the A-side for ethyl and NOELs for flavouring agents was shown by the re-
isoamyl esters and the B-side for allyl esters (Table 1). view by Munro and Kennepohl (2001), in which esti-
Adequate data were not available on allyl furoate and mated intakes were compared with the NOEL values
this compound was deferred to a later meeting because available from good quality, multi-dose toxicological
at that time the procedure did not include the thresh- studies. Most of the toxicity studies were short-term
old of toxicological concern of 1.5 !g per day on the or 90-day studies, and so may have overestimated the
B-side. Subsequent meetings considered the general NOEL that would be obtained from a chronic bioassay
threshold of 1.5 !g per day and the issue of com- on the compound or a structural analogue by a factor of
pounds such as ",#-unsaturated carbonyl compounds about three-fold (WHO, 1994). The data showed that,
and their precursors in relation to the activity of detox- with the exception of ethanol, the margins of safety
ication pathways such as oxidation and conjugation exceeded the usual 100-fold uncertainty factor; in the
with glutathione. majority of case the margins of safety were over 1000,
To date 1259 different flavouring agents have been and in many cases more than 1,000,000. This analysis
evaluated using the procedure, with the majority in supported the conclusions reached by the JECFA us-
structural class I (Table 1), and with most (979) be- ing the procedure for the safety evaluation of flavour-
ing evaluated using the A-side of the procedure (one ing agents.
compound was not evaluated in detail because it was The JECFA experience with the application of the
not related to other members of the group). Only 68 threshold of toxicological concern principle has given
compounds that were evaluated using the A-side of the rise to increased interest in its use in other risk as-
procedure had intakes that exceeded the relevant struc- sessment scenarios. Further analyses of the threshold
tural class threshold; of these 39 were of no safety con- of toxicological concern concept were presented by
cern because they were endogenous or metabolised to Kroes et al. (2000) and discussed at an international
endogenous compounds, and 29 were evaluated using workshop organised by ILSI-Europe (Barlow et al.,
toxicity data on the flavouring agent or a structurally 2001), which indicated that the approach could be
related compound. A total of 279 compounds were more widely applicable if a pre-screening procedure
evaluated using the B-side and half of these were a removed from consideration compounds that could
large and complex group of sulphides and thiol com- show bioaccumulation, genotoxicity, allergenicity or
pounds. Two compounds that were evaluated using organophosphate-like neurotoxicity. These different
the B-side were subjected to a more comprehensive aspects have been considered by an ILSI-Europe Ex-
safety assessment because their exposures exceeded pert Group and a paper (Kroes et al., 2004) prepared
the relevant threshold (JECFA, in press). In nearly all following discussions at an international workshop. A
other cases the other flavouring agents evaluated us- new decision tree is proposed in that paper in which
ing the B-side were considered not to be a safety con- compounds with structural alerts for genotoxicity are
cern based on toxicity data on the flavouring agent or considered first, and then any compound with very
a structurally related compound. Only six compounds low estimated intakes, i.e. less than 1.5 !g per day,
were considered not to be a safety concern because is removed from further detailed consideration. Such
their estimated intakes were less than 1.5 !g per day, an approach would allow assessments to concentrate
A.G. Renwick / Toxicology Letters 149 (2004) 223–234 233

on those compounds with potentially significant JECFA, 2000b. Evaluation of certain food additives and
intakes. contaminants. Fifty-third Report of the Joint FAO/WHO
Expert Committee on Food Additives. WHO Technical Report
Series, No. 896. http://www.who.int/pcs/jecfa/JECFA publi-
cations.htm.
References JECFA, 2001. Evaluation of certain food additives and
contaminants. Fifty-fifth Report of the Joint FAO/WHO
Expert Committee on Food Additives. WHO Technical Report
Barlow, S.M., Kozianowski, G., Wurtzen, G., Schlatter, J.,
Series, No. 901. http://www.who.int/pcs/jecfa/JECFA publi-
2001. Threshold of toxicological concern for chemical
cations.htm.
substances present in the diet. Food Chem. Toxicol. 39, 893–
905. JECFA, 2002a. Evaluation of certain food additives and
contaminants. Fifty-seventh Report of the Joint FAO/WHO
Cheeseman, M.A., Machuga, E.J., Bailey, A.B., 1999. A tiered
Expert Committee on Food Additives. WHO Technical Report
approach to threshold of regulation. Food Chem. Toxicol. 37,
Series, No. 909. http://www.who.int/pcs/jecfa/JECFA publi-
387–412.
cations.htm.
Cramer, G.M., Ford, R.A., Hall, R.L., 1978. Estimation of toxic
hazard—a decision tree approach. Food Cosmet. Toxicol. 16, JECFA, 2002b. Evaluation of certain food additives. Fifty-
255–276. ninth Report of the Joint FAO/WHO Expert Committee on
Food Additives. WHO Technical Report Series, No. 913.
FDA, 1993. Toxicological principles for the safety assessment
http://www.who.int/pcs/jecfa/JECFA publications.htm.
of direct food additives and color additives used in food.
“Redbook II”. Draft U.S. Food and Drug Administration. JECFA, in press. Evaluation of certain food additives. Sixty-first
Center for Food Safety and Applied Nutrition, Washington. Report of the Joint FAO/WHO Expert Committee on Food
Additives. WHO Technical Report Series.
Federal Register, 1993. Food additives; Threshold of regulation
for substances used in food-contact articles. Federal Register Kroes, R., Galli, C.L., Munro, I., Schilter, B., Tran, L.-A., Walker,
58 (195), 52719–52727. R., Würtzen, G., 2000. Threshold of toxicological concern for
Federal Register, 1995. Food additives; Threshold of regulation chemical substances present in the diet: a practical tool for
for substances used in food-contact articles. Final rule. Federal assessing the need for toxicity testing. Food Chem. Toxicol.
Register 60 (136), 36582–36594. 38, 255–312.
Fung, V.A., Barrett, J.C., Huff, J., 1995. The carcinogen bioassay Kroes, R., Muller, D., Lambe, J., Lowik, M.R.H., van Klaveren,
in perspective: application in identifying human cancer hazards. J., Kleiner, J., Massey, R., Mayer, S., Urieta, I., Verger, P.,
Environ. Health Persp. 103, 680–683. Visconti, A., 2002. Assessment of intake from the diet. Food
Gold, L.S., Slone, T.H., Bernstein, L., 1989. Summary of Chem. Toxicol. 40, 327–385.
carcinogenic potency and positivity for 492 rodent carcinogens Kroes, R., Renwick, A.G., Cheeseman, M., Kleiner, J.,
in the carcinogenic potency database. Environ. Health Persp. Mangelsdorf, I., Piersma, A., Schilter, B., Schlatter, J., van
79, 259–272. Schothorst, F., Vos, J.G., Würtzen, G., 2004. Structure-
Hall, R.L., Ford, R.A., 1999. Comparison of two methods to assess based thresholds of toxicological concern (TTC): guidance for
the intake of flavouring substances. Food Addit. Contam. 16, application to substances present at low levels in the diet. Food
481–495. Chem. Toxicol. 42, 65–83.
JECFA, 1995. Evaluation of certain food additives and Lambe, J., Cadby, P., Gibney, M., 2002. Comparison of stochastic
contaminants. Forty-fourth Report of the Joint FAO/WHO modelling of the intakes of intentionally added flavouring
Expert Committee on Food Additives. WHO Technical Report substances with theoretical added maximum daily intakes
Series, No. 859. http://www.who.int/pcs/jecfa/JECFA publi- (TAMDI) and maximized survey-derived daily intakes (MSDI).
cations.htm. Food Addit. Contam. 19, 2–14.
JECFA, 1997. Evaluation of certain food additives and Munro, I.C., 1990. Safety assessment procedures for indirect food
contaminants. Forty-sixth Report of the Joint FAO/WHO additives: an overview. Report of a workshop. Regulat. Toxicol.
Expert Committee on Food Additives. WHO Technical Report Pharmacol. 12, 2–12.
Series, No. 868. http://www.who.int/pcs/jecfa/JECFA publi- Munro, I.C., 1996. A procedure for the safety evaluation of
cations.htm. flavoring substances. Toxicological evaluation of certain food
JECFA, 1999. Evaluation of certain food additives and additives and contaminants. Prepared for the Joint FAO/WHO
contaminants. Forty-ninth Report of the Joint FAO/WHO Expert Committee on Food Additives. WHO Technical Report
Expert Committee on Food Additives. WHO Technical Report Series, Number 35, Annex 5.
Series, No. 884. http://www.who.int/pcs/jecfa/JECFA publi- Munro, I.C., Kennepohl, E., 2001. Comparison of estimated daily
cations.htm. per capita intakes of flavouring substances with no-observed-
JECFA, 2000a. Evaluation of certain food additives. Fifty- effect levels from animal studies. Food Chem. Toxicol. 39,
first Report of the Joint FAO/WHO Expert Committee on 331–354.
Food Additives. WHO Technical Report Series, No. 891. Munro, I.C., Ford, R.A., Kennepohl, E., Sprenger, J.G., 1996.
http://www.who.int/pcs/jecfa/JECFA publications.htm. Correlation of structural class with no-observed effect levels: a
234 A.G. Renwick / Toxicology Letters 149 (2004) 223–234

proposal for establishing a threshold of concern. Food Chem. Renwick, A.G., Barlow, S.M., Hertz-Picciotto, I., Boobis, A.R.,
Toxicol. 34, 829–867. Dybing, E., Edler, L., Eisenbrand, G., Greig, J.B., Kleiner,
Munro, I.C., Shubik, P., Hall, R., 1998. Principles for the safety J., Lambe, J., Müller, D.J.G., Smith, M.R., Tritscher, A.,
evaluation of flavouring substances. Food Chem. Toxicol. 36, Tuijtelaars, S., van den Brandt, P.A., Walker, R., Kroes, R.,
529–540. 2003. Risk characterisation of chemicals in food and diet. Food
Munro, I.C., Kennepohl, E., Kroes, R., 1999. A procedure for Chem. Toxicol. 41, 211–271.
the safety evaluation of flavouring substances. Food Chem. WHO, 1987. Principles for the assessment of risk to human
Toxicol. 37, 207–232. health from exposure to chemicals. Environmenal Health
Munro, I.C., Hlywka, J.J., Kennepohl, E.M., 2003. Risk asses- Criteria, 70. World Health Organization, Geneva. http://www.
sment of packaging materials. Food Addit. Contam. 19, 3– inchem.org/documents/ehc/ehc/ehc70.htm.
12. WHO, 1994. Assessing human health risks of chemicals:
Renwick, A.G., 1999. Exposure estimation, toxicological require- derivation of guidance values for health-based exposure
ments and risk assessment. In: van der Heijden, K., Younes, limits, Environmental Health Criteria, 170. World Health
M., Fishbein, L., Miller, S. (Eds.), International Food Safety Organization, Geneva. http://www.inchem.org/documents/ehc/
Handbook, Marcel Dekker, New York, pp. 59–94. ehc/ehc170.htm.

You might also like