Sage Journals: Discover world-class research

Abstract

The cancer threshold of toxicological concern (TTC) as determined by Kroes et al. in 2004 (0.15 µg/day and 1.5 µg/day at risk levels of 1 in 10⁶ or 10⁵, respectively) has been uncritically employed as a key metric in multiple regulatory guidance documents. There are numerous concerns regarding transparency and the highly conservative methodology in relation to its derivation; moreover, no formal confirmation has been undertaken by any regulatory body prior to its adoption. A recent joint report from the European Food Safety Authority and World Health Organization follows this trend, largely replicating previous conclusions and downplaying the need for a re-assessment. This view is challenged by Boobis et al. who confirm concerns regarding lack of transparency and stress that several of the assumptions and approaches used previously have been superseded by advances in knowledge; they recommend as a first step construction of a new dataset derived from the Carcinogenic Potency Database focusing on mutagenic DNA-reactive rodent carcinogens and a critical assessment of the relevance and reliability of carcinogenicity data. This type of approach is supported with two key exceptions: inclusion of data from the ToxTracker assay which provides a direct readout of DNA reactivity, and use of appropriate epidemiological data on acrylamide (AA) to determine a benchmark for human exposure to a typical DNA-reactive rodent carcinogen. It is concluded that a robust re-evaluation using an appropriate dataset and methodology is urgently needed to ensure the integrity of the cancer TTC before it is employed in its present form even more widely.

Keywords

Cancer TTC foodstuffs pharmaceutical impurities linear no-threshold (LNT) concept structural alerts for mutagenicity

Introduction

The cancer threshold of toxicological concern (TTC) is a critical element of guidance documents impacting particularly on foodstuffs and pharmaceutical impurities. In relation to the former, the latest iteration by the European Food Safety Authority (EFSA), jointly with the World Health Organization (WHO), of its position on the non-cancer and cancer TTC was published on 16th March 2016.¹ (Since EFSA was the lead contributor, citation of this report will refer only to this organization.) Consideration of the cancer TTC formed a minor yet important part of the report and EFSA’s approach showed few differences from that described by Dewhurst and Renwick in 2013.² Although various criticisms made by independent commentators are mentioned, these are ignored rather than rebutted. And the misalignment of the EFSA approach with that adopted on pharmaceutical impurities in ICH M7³ is considerably underplayed.

This article is intended to critique the derivation of the cancer TTC. Although the principal point of reference is the latest EFSA report mentioned above, this is only because no other regulatory body has provided a similar detailed justification for the derivation and adoption of the cancer TTC. An earlier article⁴ on this topic focused on EFSA’s initial position^2,5; since many of the arguments used previously still apply these have been restated (and augmented) in order to ensure completeness. In addition, there are important differences between the approaches used by EFSA and those for pharmaceuticals in relation to the application of the TTC concept. Such differences refer mainly to the categorization of compounds of concern as ‘genotoxic’ (foodstuffs) or ‘mutagenic’ (pharmaceuticals); use of structural alerts; numerical value of the TTC and lifetime versus less-than-lifetime (LTL) exposure.

Use of the cancer TTC in regulatory toxicology

EFSA has confirmed its adoption of the cancer TTC (0.15 µg/day or 0.0025 µg/kg/day – using 60 kg as an adult body weight – in relation to a compound containing a structural alert for genotoxicity) as derived by Kroes et al.⁶ (who refer to structural alerts in relation to carcinogenicity). In coming to this position, EFSA⁵ supported the methodology employed by Kroes et al. which involved linear extrapolation of cancer potency data for several hundred ‘carcinogens’ to a risk level of 1 in 10⁶. EFSA deemed the current cancer TTC value to be ‘sufficiently protective’ and considered that a re-assessment was not a priority.¹

In 2006, an industry group⁷ suggested by reference to the Kroes et al. publication (and based on a risk level of 1 in 10⁵) a cancer TTC of 1.5 µg/day as the lifetime limit for a genotoxic impurity in a pharmaceutical product or drug substance, excepting compounds in the CoC (cohort of concern) containing high-potency substances such as N-nitroso, (alkyl)azoxy and aflatoxin-like compounds. At the time, industry scientists were concerned about the potential requirement for carcinogenicity data on such impurities and perceived the TTC as a viable alternative. However, the numerical value of the TTC was accepted without any independent re-evaluation.⁸ The same value was subsequently adopted by the European Medicines Agency (EMA), Food and Drug Administration (FDA) and International Conference on Harmonization (ICH) in their guidance documents on genotoxic/mutagenic impurities in pharmaceuticals. The Kroes et al. publication was cited in the EMA and FDA guidances, but in ICH M7³ (Assessment and control of DNA-reactive (mutagenic) impurities in pharmaceuticals to limit potential carcinogenic risk), where the methodology used to derive the cancer TTC is described as ‘very conservative’ (without any further explanation), no supporting references are provided. (Note that the cancer TTC in ICH M7 is expressed only in µg/day since it is used to determine allowable concentrations of (potentially) mutagenic impurities; for example, in the case of a mutagenic impurity in a drug used at a maximum daily dose of 150 mg for the treatment of a chronic condition, application of the default lifetime TTC produces an impurity limit of 10 ppm.)

Criticisms of the TTC derivation

A number of criticisms of the Kroes et al. methodology and the validity of the conclusions have been made by Delaney,⁹ Snodin and McCrossen¹⁰ and Snodin.⁴ In summary, these are as follows:

Skewed dataset(s) derived from the Carcinogenic Potency Database (CPDB) based on a preponderance of compounds expected to be carcinogenic/mutagenic;

Use of the lowest statistically significant TD₅₀ (a cancer-potency metric) instead of the default harmonic-mean TD₅₀ leads to the phenomenon of false carcinogens and often overestimates carcinogenic potency. (Alternatively the most reliable/appropriate TD₅₀s could be used as described in ICH M7 R1¹¹)

Linear extrapolation of cancer potency values for non-mutagenic carcinogens, which are generally considered to be thresholded, is inappropriate.

A further issue in relation to point 3 is the growing concern about the use of the linear no-threshold (LNT) concept for assessment of carcinogenic risk (initially for radiation and subsequently adopted for chemical carcinogens) in that there appears to be clear evidence of misrepresentation and lack of transparency dating back to its development and promulgation in the 1950s and 1960s.^12,13 Discontinuities between the effects of chemicals at high and low doses often make simple linear extrapolation unrealistic. Since multiple mechanisms operate in mammalian systems, the assumption of a simple dose–response relationship and a single outcome over the low- to high-dose range is likely to be highly misleading.

Skewed dataset for determination of cancer TTC

The dataset used by Kroes et al.,⁶ stated to contain 730 compounds, for the determination of the cancer TTC was non-transparent, which still remains the case. Various unsuccessful attempts have been made to retrieve the dataset, described as ‘no longer available’.¹⁴ Several of the co-authors of the 2004 article appear to remain professionally active and it seems strange that none has come forward to clarify the position and/or release a copy of the dataset. However, it can be reasonably assumed¹⁰ that the vast majority of the compounds (claimed to be 709; 706 listed) were those employed to support an earlier evaluation by Cheeseman et al.¹⁵ using closely similar inclusion and cancer-potency criteria.

Cheeseman et al. commented on the skewed composition of their dataset as follows:

Another conservativism is the nature of the CPDB itself as a basis for establishment of a regulatory threshold. The overall percentage of positive carcinogens in the CPDB is approximately 54%, which is comparable to the level of 52% reported for the smaller NCI/NTP (National Cancer Institute/National Toxicology Program) subset of carcinogenicity studies. Huff et al. (1995) evaluated several aspects of the NCI/NTP database, including the basis for nomination of the substances for study. In general, substances were nominated either because they were suspected of carcinogenic activity based on structural concerns or they represented substances of potentially high exposure. Substances nominated as suspected carcinogens accounted for 86% of those substances positive in at least one species in the NCI/NTP database. These compounds represent 90% of compounds positive in more than one species. This high rate of positivity is indicative of a database that is (justifiably) skewed towards expected carcinogens. In contrast, Huff et al. (1995) postulated that an expectation of a reasonable range of human carcinogenicity is likely to be no higher than 5-10% of chemicals in production. From this, we can expect that a regulatory threshold based on the CPDB or the NCI/NTP subset would be anywhere from fivefold to 10-fold more conservative than a threshold derived from a subset more representative of the larger population of commercially produced substances.

On the other hand, the EFSA report endeavours to negate the notion that the CPDB-derived dataset is skewed, relying on a chemoinformatics-based chemical-space analysis using physicochemical and molecular-structure descriptors.¹⁶ (The minimum oral TD₅₀ value for each chemical was determined as per Cheeseman et al. in order to construct a dataset of 653 compounds. Those studies considered negative by authors were still included in this version if the studies met the Cheeseman criteria. The metric employed was log (1/TD₅₀) where the TD₅₀ is reported as mol/kg/day, which is inconsistent with the Cheeseman and Kroes assessments based on mg/kg/day. Using a metric based on molar doses changes potency ranking to some extent, increasing the relative potency of high-molecular-weight compounds such as aflatoxins and azo dyes, and decreasing that of low-molecular-weight compounds such as acrylonitrile.)

EFSA asserts that the dataset is broadly representative of the ‘world of chemicals’ and comments as follows:

The term “world of chemicals” is a poorly defined term and can be interpreted in various ways. In the present context the term is meant to address the question whether the structure(s) of a chemical under consideration is represented by the chemicals in the database used to derive the respective TTC value.

This definition is based only on chemical structures and excludes issues related to the relative proportions of high-, mid- and low-potency compounds in the dataset. By contrast, Cheeseman et al. freely acknowledge the justifiably skewed nature of the cancer TTC dataset and of the underlying CPDB dataset – currently containing 1547 chemicals – which over the years have been shown consistently to contain approximately 50% of carcinogens, whereas estimates of the proportion of carcinogens among commercially produced chemicals are 5–10% (see above and Fung et al.¹⁷). In order to support their ‘world of chemicals’ argument, EFSA relies on a technique that employs potentially inappropriate surrogate markers based on chemometric parameters. In order to validate this kind of approach, a sensitivity analysis could be performed involving further chemometric evaluations using deliberately skewed datasets containing an increasing proportion of potent carcinogens of various structural types in order to assess whether there is a point at which a difference from an appropriate reference dataset can be detected. For example, if the test dataset were to contain solely structures from the ‘cohort of concern’(CoC) (which comprises several classes of high-potency mutagenic carcinogens) would it still be considered to be representative of the ‘world of chemicals’?

The problem of selection bias is confirmed by Boobis et al.¹⁸:

these databases are impacted by a large selection bias because those substances tested were predominantly already suspected of being carcinogens, based on structural features or toxicological effects. Indeed, Fung et al. (1995) reported in their analysis that whilst of 400 substances tested in cancer bioassays by NCI/NTP, 210 were positive, of these, 181 had been selected for the testing program based on suspected carcinogenicity, while only 29 positive substances (7% of 400) had been chosen due to high exposure potential.

The same authors also highlight the problem that many of the studies used doses greatly exceeding the maximum tolerated dose which probably meant that some of the tumour findings were secondary to other toxicity.

Regarding the Cheeseman et al. dataset, if various compound types are eliminated from further consideration – CoC; CoC substances, endocrine disruptors, organophosphorus compounds, heavy-metal compounds and low/moderate-potency carcinogens (TD₅₀ > 50 mg/kg/day) – only around 150 critical compounds remain of which up to 20% appear to be non-mutagens (based on data presented by Snodin and McCrossen¹⁰).

Lowest statistically significant TD₅₀ versus default harmonic-mean TD₅₀

EFSA released a scientific opinion (dated 1 March 2016) on approximately 140 substances from the CPDB that are of particular concern.¹⁹ A high proportion of the compounds involved can be classified as belonging to the CoC; for example, around 50% are N-nitroso compounds. It is noteworthy that harmonic-mean TD₅₀s were employed throughout, and in general, the lowest statistically significant TD₅₀s cited in Cheeseman et al. are distinctly lower. The latter values are 50–90% of the default value for several non-mutagenic pharmaceuticals/pesticides such as dieldrin, enovid, kepone, methimazole, phenesterin and heptachlor. On the one hand, EFSA is content to employ harmonic-mean TD₅₀s for this exercise, whereas the (presumed) use of the lowest statistically significant TD₅₀ by Kroes et al.⁶ is supported by EFSA in relation to derivation of the cancer TTC. Concerning false carcinogens, an important example is carbamate esters for which Cheeseman et al. list around 18 compounds, all claimed to be carcinogens, whereas the CPDB indicates that all but five show no positive test. In addition, the majority are non-mutagenic. (Some carbamates that are reported to be mutagenic may reflect the presence of mutagenic impurities in technical-grade test materials.) A further example is acrolein which is listed in the CPDB as showing no positive test in either rats or mice²⁰; on the other hand, a TD₅₀ of 14.4 mg/kg/day is cited by Cheeseman et al.¹⁵ Methyl methanesulfonate and captan are two examples where the use of the lowest statistically significant TD₅₀ (as listed by Cheeseman et al.) grossly overestimates carcinogenic potency (based on the geometric-mean value).

Evaluation of non-mutagenic carcinogens

Gold et al.²¹ determined that for compounds in the CPDB at the time a high proportion of both mutagens and non-mutagens were carcinogenic, but the proportion was higher for mutagens (75%) than non-mutagens (45%).

A statement confirming the generally thresholded nature of non-mutagenic carcinogens is provided by the EFSA in their 2016 report,¹ as follows:

Carcinogens which are not directly DNA-reactive often have a threshold mode of action and, in general, NOAELs for these are in the same range or higher than NOAELs for other types of toxicity. Thus, EFSA (2012) concluded that TTC values that are higher than the value of 0.15 μg/person/day are appropriate for any chemical where the weight of evidence for DNA reactivity is negative. The expert group concurred with this statement.

On the other hand, the EFSA seems minded to ignore the potential impact of non-mutagenic/non-DNA-reactive carcinogens in the TTC dataset:

Although further expansion of the TTC cancer dataset and an updated carcinogenicity potency database is desirable, it is not expected that the overall threshold for carcinogenicity based on linear extrapolation for genotoxic carcinogens would significantly change.

It should be remembered that the value of the cancer TTC relies predominantly on TD₅₀ data for a limited number of compounds (estimated to be around 150 as mentioned above). And so the presence of relatively few non-mutagenic substances (estimated to be 20%) in this critical subset could create significant bias.

Approaches for pharmaceutical impurities and LTL exposure

The LTL concept is described by the EFSA as follows:

The current uses of TTC, with the exception of impurities in pharmaceuticals and possibly consumer products, are primarily for lifetime human exposure assessments. A number of proposals have been put forward for adjusting the TTC value for chemicals with a structural alert for genotoxicity for shorter than chronic durations of exposure (Müller et al, 2006, Humfrey 2007, Felter et al, 2009, Galloway et al. 2013), mainly in the context of impurities in pharmaceuticals and consumer products. The proposals made are mainly based on Haber’s law for cancer risk assessment, i.e. that higher exposures for shorter periods of time are essentially equivalent to lower exposures for longer periods of time. The EFSA (2012a) was not confident about the general applicability of these proposals. It therefore recommended that the issue of less than chronic exposure should be addressed on a case-by-case basis. This could be done for example by considering the margin between the appropriate TTC value (without any adjustment for duration of exposure) and the estimated dietary exposure.

The above commentary is considered to be an inaccurate and incomplete description of the use of structural alerts in the context of pharmaceutical impurities, for the following reasons:

There is no mention of the ICH M7 consensus guideline on mutagenic impurities³;

The EFSA statement mentions genotoxicity, whereas in ICH M7, structural alerts are interpreted in the context of mutagenicity;

The determination of bacterial mutagenic potential in ICH M7 is a sophisticated and multi-layered procedure involving inter alia;

In silico evaluation using two complementary systems (rule- and statistical-based);

Follow-up using expert assessment;

Use of read-across from appropriate reference compounds (e.g. evaluating the drug substance and impurities for the presence of similar structural moieties);

Performing an Ames’ assay, as necessary;

For human pharmaceuticals data from mammalian-cell assays are discounted if the impurity is predicted with confidence to be non-mutagenic and/or is an experimental non-mutagen – not the case veterinary;

If an impurity with a conventional structural alert for bacterial mutagenicity is shown by either in silico prediction or experimental means to be a non-mutagen, it is treated as a normal (non-mutagenic) impurity with no structural alert;

Compound-specific limits can be determined for mutagenic impurities mainly based on the determination of an in vivo threshold for mutagenic effects²² establishing that the mechanism of carcinogenicity is unrelated to genotoxicity, or linear extrapolation of carcinogenic potency data.¹¹ Some examples are shown in Table 1.

Table 1.

Examples of compound-specific limits for mutagenic impurities in pharmaceuticals.

Compound	Method of determination	Lifetime limit (µg/day)	Reference
Benzyl chloride	Linear extrapolation of TD₅₀	41	¹¹
1-Chloro-4-nitrobenzene	Linear extrapolation of TD₅₀	117	¹¹
Chloroethane	Linear extrapolation of TD₅₀	1810	¹¹
Chloromethane	Linear extrapolation of TD₅₀	1360	¹¹
Aniline	Thresholded mechanism of carcinogenesis	720	¹¹
Hydroxylamine	Thresholded mechanism of carcinogenesis	2	¹¹
Methyl methanesulfonate	Linear extrapolation of TD₅₀	31.8	²²
Ethyl methanesulfonate	Threshold for in vivo mutagenicity	104	²²

Thus, EFSA’s characterization of the ICH M7 approach as applying broadly to ‘chemicals with a structural alert for genotoxicity’ is extremely misleading. ICH M7 indicates: ‘The existence of impurity structural alerts alone is considered insufficient to trigger follow-up measures, unless it is a structure in the cohort of concern’. Whereas, EFSA relies on conventional structural alerts for genotoxicity,²³ which are often inappropriate and significantly over-predictive, the precise targeting of the potential for bacterial mutagenicity using in silico and other techniques enables a number of conventional alerts to be discounted. One such alert is the α, β-unsaturated ketone moiety,¹⁰ and a key example is mesityl oxide (4-methylpent-3-en-2-one, CAS no 141-79-7; a potential minor contaminant in acetone) which was used frequently as a source of regulatory-agency deficiency questions if acetone was used as a solvent in the synthesis of a drug substance. In fact, mesityl oxide tests negative for mutagenicity in five strains of Salmonella typhimurium and in the in vitro micronucleus assay using human blood peripheral lymphocytes.²⁴

Structural alerts and in silico prediction of mutagenic potential are also discussed by Boobis et al.¹⁸ who conclude that ‘It is thus critical in the application of the TTC approach that those compounds likely to be mutagenic via a DNA-reactive mode action can be reliably identified on the basis of structure alone’. As noted above, experience with pharmaceutical impurities indicates that employing structural alerts alone for predicting mutagenic potential can be erroneous in many cases. On the other hand, use of historical data and (commercial) in silico systems supported by large datasets provides the most reliable predictions, particularly when combined with an expert assessment (e.g. evaluating mode of action (MOA), structural analogues, data reliability and overall weight of evidence).²⁵ However, both structural alerts and in silico QSAR systems are reported to be over-predictive for compounds containing bulky or electron-withdrawing substituents because an alert is triggered when a particular toxicophore is present irrespective of the structural environment surrounding the toxicophore.²⁶

As regards the LTL concept, this is based on Haber’s rule, a fundamental concept in toxicology, where concentration (C) × time (T) = a constant (k), the carcinogenic effect being related to both dose and duration of exposure. The LTL levels proposed in ICH M7 are significantly lower than values calculated using Haber’s rule and are also in compliance with a 10⁻⁶ cancer risk level if treatment durations are no longer than 6 months.³ LTL multiples of the generic lifetime limit (1.5 µg/day) or compound-specific lifetime limit are 80, 13.3 and 6.7 for exposure durations of 1 month, 1–12 months and 1–10 years, respectively. Thus, the limit for short-term (≤ 1 month) exposure to hydroxylamine (Table 1) would be 160 µg/day. Haber’s rule has been used extensively in the context of Registration, Evaluation, Authorisation and Restriction of Chemicals²⁷ and in regulations on the control of substances hazardous to health, in which both frequency and duration of exposure are considered to be critical.²⁸

The concept of a threshold for mutagenic carcinogens is a closely related issue for which there are a number of examples such as ethyl methanesulfonate²⁹ and AA (acrylamide, whose metabolite glycidamide is a direct-active bacterial mutagen³⁰). In its most recent review³¹ of AA, EFSA reported (inter alia) that:

Mean and 95th percentile dietary AA exposures across surveys and age groups were estimated at 0.4–1.9 μg/kg body weight (b.w.) per day and 0.6–3.4 μg/kg b.w. per day, respectively.

Although the epidemiological associations have not demonstrated AA to be a human carcinogen, the margins of exposure indicate a concern for neoplastic effects based on animal evidence.

Numerous epidemiological studies have been undertaken on AA and in a 2012 review Lipworth et al.³² state:

we found no consistent or credible evidence that dietary acrylamide increases the risk of any type of cancer in humans, either overall or among non-smokers. In particular, the collective evidence suggests that a high level of dietary acrylamide intake is not a risk factor for breast, endometrial, or ovarian cancers, which have generated particular interest because of a conjectured hormonal mechanism of acrylamide.

In a follow-up comment, Lipworth et al.³³ emphasize that occupational exposures to AA, which are much higher than dietary levels, show no association with an increased risk of cancer. More recent epidemiological reports have detected no association with dietary AA and cancer.^34
–36 Whether or not dietary AA constitutes an unambiguous carcinogenic risk is currently uncertain; however, AA is considered to be a rare example of genotoxic rodent carcinogen that has been subjected to an extremely comprehensive epidemiological evaluation and found to show no overall association with an increased cancer risk. The average dietary exposure of AA is approximately 0.5 µg/kg/day, 200 times the EFSA cancer TTC of 0.0025 µg/kg/day, which does seem to raise concerns over the human relevance of the latter value based entirely on extrapolation of rodent carcinogenic potency data. (According to the CPDB, AA is categorized as a bacterial mutagen and the lowest statistically significant TD₅₀ is 1.75 mg/kg/day,³⁷ linear extrapolation of which using a factor of 10⁶ yields a compound-specific value of 0.00175 µg/kg/day, closely similar to the EFSA cancer TTC.) Other compounds or groups of compounds such as aflatoxins, on which extensive epidemiological studies have been conducted, could be evaluated in relation to assessing the validity of the cancer TTC and/or the reliability of linear extrapolation of animal cancer potency data.

Proposals for re-assessment of the cancer TTC by Boobis et al.¹⁸

Following a detailed commentary on the current status of the cancer TTC, Boobis et al. conclude:

it is apparent that the current TTC values for substances assumed to be DNA-reactive mutagens and carcinogens could be questioned, for a number of reasons. These values are likely to be conservative.…From a scientific perspective, there is a clear case to be made for the re-assessment of these values. While existing values appear to be adequately protective of human health, a number of assumptions and approaches used in their original derivation have been superseded by advances in knowledge. Hence, to ensure a robust, transparent basis for all aspects of the TTC approach, re-assessment of these values is timely. In addition, restrictions on the use of chemicals should be commensurate with their risk, which may not be the case with the application of the current values.

The final sentence can be interpreted as strongly implying that the current TTC values may be over-conservative thus leading to unnecessarily over-restrictive controls on DNA-reactive mutagens and carcinogens. Table 2 summarizes the principal requirements described by Boobis et al. for construction of a dataset focused on potency data (from the CPDB) for ‘DNA-reactive mutagenic carcinogens’. The authors point out that while it would be valuable to update and expand the CPDB there are likely to be relatively few new reports on carcinogenicity bioassays for DNA-reactive mutagens (since public health and commercial considerations militate against the development of such compounds).

Table 2.

Principal requirements for constructing a suitable dataset for re-evaluation of the cancer TTC¹⁸.

Issue	Proposal
Goal	To construct a curated, transparent dataset of carcinogens using inclusion and exclusion criteria reflecting current knowledge and weight of evidence for genotoxicity and carcinogenicity. Distinguishing substances that are carcinogenic by a DNA-reactive mutagenic mode of action (MOA) from those that clearly act by a non-genotoxic MOA is a critical requirement
Evaluation of assumptions in previous analyses	Among assumptions that have been made in the development of this TTC value are (1) the compounds in the CPDB are representative of the world of carcinogens, (2) there are sufficient compounds to obtain a reliable estimate of the distribution of their PODs (points of departure), (3) carcinogens have a linear dose–response curve, (4) the carcinogenic response observed in laboratory animals is relevant to humans, (5) the acceptable excess risk for such compounds is 1 × 10⁻⁶
Definition of DNA-reactive mutagenic carcinogens	DNA-reactive mutagens are substances that can react with and modify DNA directly, either as the parent chemical or in the form of a metabolite. Chemicals are defined as carcinogenic if they induce tumours, increase tumour incidence and/or malignancy or shorten the time to tumour occurrence, relative to controls. Genotoxic chemicals that can cause cancer as a result of their genotoxicity are typically referred to as genotoxic carcinogens. More specifically, where this is a consequence of DNA-reactive mutagenicity, such compounds have been referred to as DNA-reactive mutagenic carcinogens
Updating CPDB	While it would be valuable to expand the CPDB, for the present exercise, it is unlikely that very many suitable new compounds would be available
Importance of mechanisms	Identification of mutagenic/genotoxic carcinogens that have a thresholded MOA
Evaluation of genotoxicity data	Strength of evidence assessed based on positive bacterial reverse mutation assay plus data from other genotoxicity assays. In vivo genotoxicity in the target species and target tissue for carcinogenicity is considered to provide considerable weight in determining whether a compound acts by a DNA-reactive mutagenic MOA
Evaluation of carcinogenicity data	Evaluation of data from CPDB in relation to various criteria including: study relevance and reliability (Klimisch score), route of exposure, tumour findings relevant to humans, study duration and number of surviving animals (minimum number 10 and 20/sex/group for positive and negative studies, respectively), MTD achieved but not exceeded, exclusion of data from positive single-dose-group studies

Discussion and Conclusions

The three key criticisms of the derivation of the cancer TTC are considered to remain unanswered.

Concerning the issue of the skewed dataset, Cheeseman et al.,¹⁵ whose analysis employed a closely similar dataset to that used by Kroes et al.⁶, clearly state that their dataset is (justifiably) skewed to such an extent that any regulatory threshold would be 5- to 10-fold more conservative than that derived from a subset more representative of the larger population of commercially produced substances. The latter figure appears to be based on comparison of the proportion of carcinogens in the CPDB (50%) with that estimated in chemicals of commerce (5–10%). Thus, a random selection of 1500 substances from those used in commerce would be expected, on average, to contain 75–150 carcinogens most probably with potencies significantly lower than CPDB carcinogens which were for the most part selected on the basis of being likely (potent) carcinogens. EFSA has never mentioned this kind of evaluation and instead attempts to construct a contradictory argument based on a ‘world of chemicals’ chemoinformatics concept.

Regarding the use of the lowest statistically significant TD₅₀s in the derivation of the TTC by Kroes et al., which is defended by EFSA, this stance appears to have been reversed by evidence from a recent EFSA report that uses harmonic-mean TD₅₀s throughout.¹⁹ Although EFSA has confirmed its position that linear extrapolation of carcinogenic potency data is inappropriate for non-mutagenic carcinogens, so far it has ignored the presence/impact of such compounds in the CPDB dataset. Over the last decade or so, there have been multiple missed opportunities to perform an ab initio evaluation of a suitable dataset using appropriate inclusion and evaluation criteria, and the outcome of the non-transparent and arguably flawed evaluation by Kroes et al. has been accepted by EFSA and other regulatory bodies. The possibility cannot be discounted that there has been a general policy of avoiding any re-evaluation for fear of a less conservative outcome. This is in marked contrast to the use of transgenic mouse models in carcinogenicity bioassays which, before their widespread acceptance, were subjected to a detailed review by a working party of the EMA.³⁸ A joint report by three EU expert committees (SCHER, SCENIHR, SCCP) was also somewhat critical of the cancer TTC in relation to being based on worst-case assumptions, assuming no threshold for carcinogenicity and databases requiring further development and validation.³⁹

A more general point on linear extrapolation down to extremely low risk levels is made by Cheeseman et al.¹⁵: ‘the procedure of linear extrapolation to low dose is highly conservative and may exaggerate risk by as much as two orders of magnitude. This may be particularly true for non-genotoxic carcinogens’. These remarks about significant overconservatism are consistent with recent challenges to the LNT concept and its adoption by regulatory agencies.^12,13 Three of the four criticisms mentioned in previous sections (skewed dataset, lumping together of genotoxic and non-genotoxic carcinogens and overconservatism of associated with the LNT concept) were openly acknowledged by Cheeseman et al. And so it’s alarming that these issues have not been evaluated by regulatory agencies.

As noted by Galloway⁴⁰ in the United States, an individual has a 40% chance of developing cancer over a lifetime. Exposure to a mutagen/potential carcinogen at a level estimated to increase risk by 1 in 10⁶ would elevate the individual cancer risk from 0.400000 to 0.400001.

Overall, the lifetime cancer TTC is likely to be over-conservative by possibly 2 orders of magnitude. Moreover, deriving a TTC metric accurate to two significant figures (i.e. 0.15 or 1.5 µg/day) seems extremely implausible given that the procedure involved linear extrapolation of a highly variable dataset of carcinogenicity potency data over 5 or 6 orders of magnitude. Alternative cancer-potency metrics such as the T25^41
–43 (which is much simpler to determine than the TD₅₀ and so would most likely increase the size of the dataset) could be employed as mentioned by EFSA.¹ (Although these two metrics are reasonably well correlated,⁴¹ they appear to be sufficiently different to negate the determination of a TTC value to two significant figures.) Anyone familiar with pharmaceutical regulatory affairs would most likely be able to quote the value employed for lifetime exposure to a mutagenic impurity (1.5 µg/day), but based on the author’s experience relatively few have any clear notion of its derivation. For example, a 2011 article⁴⁴ asserted that the cancer TTC dataset was obtained from the Registry of Toxic Effects of Chemical Substances rather than the CPDB. It’s therefore unsurprising that a limited number of scientists have taken the time to delve into the somewhat opaque background to the cancer TTC.

EFSA’s concerns about use of the LTL principle are understandable to some extent given that the possibility of short-term exposure is much more limited for foodstuffs compared to pharmaceuticals. On the other hand, EFSA has not provided any rational argument and evidence to support its position and, moreover, the flow chart in its 2016 report¹ appears to indicate an increased limit of 1.5 µg/day for genotoxic compounds when the duration of exposure does not exceed 12 months. Two further considerations are:

The highest LTL multiple mentioned in ICH M7 is 80 in relation to exposure duration of 1 month, which is lower than the estimated extent of overconservatism of the lifetime TTC.

Research on smoking-related mortality shows a dramatic beneficial effect of early cessation. A report on one million women smokers indicated that stopping before age 40 years avoids more than 90% of the excess mortality caused by continued smoking; stopping before age 30 years avoids more than 97% of excess mortality.⁴⁵

Given the foregoing, the position of EFSA on the cancer TTC is considered to be non-robust when subjected to a critical analysis of relevant data and regulatory precedents. Furthermore, there appears to be no example of any regulatory body, including EFSA, adopting the cancer TTC only after performing an independent re-assessment in order to accommodate the various criticisms of the Kroes et al. derivation. This state of affairs could be attributed to a pattern of confirmation bias presumably because no agency would want to put itself in a position of challenging the current consensus, which ignores the general philosophy that science advances in an iterative fashion by deliberately challenging hypotheses and poorly established concepts.

The cancer TTC has gradually gathered more ‘superstructure’ as it has been applied or proposed for application in areas other than foodstuffs and pharmaceuticals, such as food contact materials, flavouring substances, herbal products, cosmetic ingredients and personal and household care products.³⁹ Recent examples are the EMA guidance on setting health-based exposure limits⁴⁶ and draft guidance on mutagenic impurities in veterinary medicines.⁴⁷ It would be a great pity and a major disservice to toxicological risk assessment if the cancer TTC as proposed by Kroes et al. were allowed to continue on its current trajectory. However, ILSI (International Life Sciences Institute) Europe is sponsoring an initiative on a re-evaluation as described by Boobis et al.¹⁸ In order to initiate this process, the authors believe that the principal requirement is the construction of a fully curated, publicly available dataset on genotoxic and non-genotoxic compounds using predetermined inclusion and exclusion criteria. In addition, assessment of the relevance of tumours for human health, dose–response and MOA for tumours detected in rodent carcinogenicity bioassays are important issues. Distinguishing compounds that are carcinogenic by a DNA-reactive mutagenic MOA from those that act via a non-genotoxic MOA is a critical consideration. While such an exercise is to be greatly encouraged, there may be difficulties in harmonizing this kind of approach and that in ICH M7 which employs bacterial mutagenicity as a surrogate for DNA-reactivity and excludes use of data from a mammalian-cell assay such as the in vitro micronucleus test.⁴⁸ There are often potential issues on deciding whether a particular compound should be tagged as a ‘mutagenic carcinogen’ or ‘genotoxic carcinogen’, and Boobis et al.¹⁸ focus on carcinogens that have a DNA-reactive mutagenic MOA (determined on the basis of a positive Ames’ test plus data from other genotoxicity assays because 20–30% of Ames’-positive compounds are reported to test negative for carcinogenicity). Unfortunately, there is no mention of the newly developed ToxTracker assay⁴⁹ which has the ability to provide a direct readout of DNA-reactivity and so should be superior to most other genotoxicity assays in providing MOA data. (Clearly weight-of-evidence evaluations will be critical across the board particularly in relation to mechanism and human relevance of rodent carcinogenesis^50
–52). An additional concern is that the initiative by Boobis et al. mentions no role for epidemiological evidence although it is acknowledged that the enzymes involved in producing DNA-reactive metabolites are frequently less active in humans than in rodents. As described previously, AA is an example of a potent genotoxic carcinogen in rodents for which extremely comprehensive epidemiological studies indicate no overall association with an increased cancer risk, and thus, the current human exposure to AA could provide a benchmark and ‘reality check’ impacting on determination of the cancer TTC.

In their derivation of the cancer TTC, Kroes et al. show that 3.9% (17 out of 434; Table 3) of compounds (i.e. with exclusion of CoC (cohort of concern) compounds, heavy-metal-containing compounds, highly chlorinated compounds, organophosphorus compounds, steroids and tetrahalogenated dibenzodioxins and dibenzofurans) possess TD₅₀ values < 1.25 mg/kg/day (corresponding to an exposure less than 1.5 µg/day using linear extrapolation of 1 in 10⁵). Presently, this small group of potent carcinogens has been effectively ignored in relation to the derivation of the lifetime TTC of 1.5 µg/day for pharmaceuticals. Consequently, based on this kind of issue and regulatory pragmatism retention of the current value of the TTC has been advocated, while accepting that the Kroes et al. derivation has many deficiencies.⁵³ Since no information is available in the public domain on the precise makeup of the dataset employed by Kroes et al., it is not possible to identify with certainty the 17 high-potency compounds. However, using the current version of the CPDB 14 mutagenic (Ames’-positive) compounds have been identified with (oral) TD₅₀s < 1.25 mg/kg/day (Table 4). Although clearly not a perfect match with the 17 high-potency compounds identified by Kroes et al., this group of compounds is considered sufficiently representative for current assessment purposes. Data from Table 4 indicate that the potency metrics for six compounds are based on studies with a single dose level and so would be excluded from any re-evaluation of the TTC using criteria set out by Boobis et al.¹⁸ Thus, on the basis of this initial analysis only 8/434 high-potency compounds (1.8%) remain, and there could be further attrition if other criteria shown in Table 2 were applied. It is believed that these ‘outlier’ compounds are likely to be of negligible significance in any re-evaluation of the cancer TTC.

Table 3.

Numbers and fractions of compounds in different structural categories (excluding CoC compounds and other categories^a) estimated to produce a risk ≥ 1 in 10⁶ at different intake levels^b.

Structural category	0.15 µg/day (fraction)^c	1.5 µg/day (fraction)^c	Total compounds/structural category
Aromatic amines	5 (0.03)	51 (0.31)	162
Aromatic nitro compounds	2 (0.06)	8 (0.24)	33
Azo compounds	0 (0)	9 (0.5)	18
Benzidine derivatives	2 (0.14)	6 (0.43)	14
Carbamates	0 (0)	8 (0.4)	20
Hydrazines	2 (0.04)	30 (0.53)	57
Miscellaneous Ashby alerts	2 (0.05)	5 (0.12)	41
α-Nitrofuryl compounds	1 (0.03)	16 (0.17)	34
Strained rings	1 (0.07)	9 (0.60)	15
Vinyl containing compounds	2 (0.05)	13 (0.33)	40

^aHeavy-metal-containing compounds, highly chlorinated compounds, organophosphorus compounds, steroids and tetrahalogenated dibenzodioxins and dibenzofurans.

^bSimplified version of Table 1 in Kroes et al.⁶ (oral route of administration).

^cCorresponding TD₅₀ s for the two intake levels are: < 1.25 and 1.25–12.5 mg/kg/day.

Fraction = compounds in particular intake category divided by total number in structural category; numbers and fractions are shown cumulatively.

Table 4.

Data from CPDB on mutagenic compounds in different structural categories with high carcinogenic potency^a.

Structural category	Examples	CPDB TD₅₀ mg/kg/day^b	Comments
Aromatic amines	None	–	–
Aromatic nitro compounds	2,6-Dinitrotoluene⁵⁴	0.292	TO = liver; two studies with two dose levels in each; rat only
Azo compounds	None	–	–
Benzidine derivatives	3,3′-Dimethylbenzidine.2HCl⁵⁵	0.629	Multiple TOs in the rat; one set of studies with up to six dose levels
Benzidine derivatives	3,3′-Dimethoxybenzidine.2HCl⁵⁶	1.04	Multiple TOs in the rat; two sets of studies with up to six dose levels
Carbamates	None	–	–
Hydrazines	1,2-Dimethylhydrazine.2HCl⁵⁷	0.114	TO = lung in mouse and liver in hamster; no rat study; two sets of studies with only one dose level in each
	2-Hydroxyethylhydrazine⁵⁸	0.397	TO = liver in male mouse; no positive test in females or in rats or hamsters; three sets of studies with only one dose level in each
	Hydrazine⁵⁹	0.613	TO = liver, lung (rat); lung (mouse). Six sets of studies with up to four dose levels
Miscellaneous Ashby alerts	Chlorambucil⁶⁰	0.133	TO = lung (mouse); slightly less potent in rat with different TOs. Two sets of studies with up to two dose levels
	1,2-Dibromo-3-chloropropane⁶¹	0.259	TO = lung, stomach (rat). Multiple studies but with no more than two dose levels
	Benzotrichloride⁶²	0.396	TO = lung, stomach (mouse). No rat test. One study with four dose levels
	1,2,3-Trichloropropane⁶³	0.875	Multiple TOs; slightly more potent in mouse. Multiple studies with three dose levels
α-Nitrofuryl compounds	2-(2,2-Dimethylhydrazino)-4-(5-nitro-2-furyl)thiazole⁶⁴	0.41	TO = haemopoietic system and mammary gland. Rat only. One study with one dose level
α-Nitrofuryl compounds	2-Amino-5-(5-nitro-2-furyl)-1,3,4-thiadiazole⁶⁵	0.662	TO = kidney, lung, mammary gland. Female rat only. One study with one dose level
Strained rings	Ethylene imine⁶⁶	0.377	TO = liver, lung (mouse only). One study with one dose level
Strained rings	β-Propiolactone⁶⁷	1.24	TO = stomach (mouse and female rat). No test in male rat. Slightly more potent in mouse. Three studies with single dose levels
Vinyl containing compounds	None	–	–

^aTD₅₀ < 1.25 mg/kg/day; TO = (main) target organ(s).

^bGeometric mean TD₅₀ in rat or mouse, whichever is lower.

Several lines of evidence support the notion that the current values of the cancer TTC values for foodstuffs and pharmaceuticals are over-conservative and not commensurate with the actual risk to human health posed by DNA-reactive mutagenic carcinogens. It is concluded that there is a strong case for re-evaluation of the cancer TTC, which is supported by Boobis et al.¹⁸ who have developed a detailed strategy for this purpose. While the Boobis et al. approach is endorsed in general, two modifications are suggested: use of the ToxTracker assay for determination of DNA-reactivity and incorporation of epidemiological data on compounds such as AA and aflatoxins.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

EFSA (European Food Safety Authority) and WHO (World Health Organization), 2016. Review of the threshold of toxicological concern (TTC) approach and development of new TTC decision tree. http://onlinelibrary.wiley.com/doi/10.2903/sp.efsa.2016.EN-1006/pdf (accessed 2 October 2017).

Dewhurst

Renwick

. Evaluation of the threshold of toxicological concern (TTC) – challenges and approaches. Regul Toxicol Pharmacol 2013; 65(1): 168–177.

ICH M7: Assessment and control of DNA reactive (mutagenic) impurities in pharmaceuticals to limit potential carcinogenic risk, http://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Multidisciplinary/M7/M7_R1_Addendum_Step_4_31Mar2017.pdf (accessed 2 October 2017).

Snodin

DJ.

Regulatory risk assessments: is there a need to reduce uncertainty and enhance robustness?

Hum Exp Toxicol 2015; 34(12): 1258–1271.

Scientific opinion on exploring options for providing advice about possible human health risks based on the concept of threshold of toxicological concern (TTC). EFSA J 2012; 10(7): 2750. http://www.efsa.europa.eu/en/efsajournal/pub/2750.htm (accessed 3 October 2017).

Kroes

Renwick

Cheeseman

. European branch of the International Life Sciences Institute. Structure-based thresholds of toxicological concern (TTC): guidance for application to substances present at low levels in the diet. Food Chem Toxicol 2004; 42(1): 65–83.

Müller

Mauthe

Riley

. A rationale for determining, testing, and controlling specific impurities in pharmaceuticals that possess potential for genotoxicity. Regul Toxicol Pharmacol 2006; 44(3): 198–211.

Humfrey

, personal communication.

Delaney

. An impact analysis of the application of the threshold of toxicological concern concept to pharmaceuticals. Regul Toxicol Pharmacol 2007; 49(2): 107–124.

10.

Snodin

McCrossen

. Mutagenic impurities in pharmaceuticals: a critique of the derivation of the cancer TTC (threshold of toxicological concern) and recommendations for structural-class-based limits. Regul Toxicol Pharmacol 2013; 67(2): 299–316.

11.

ICH M7 (R1) Step 2. Application of the principles of the ICH M7 guideline to calculation of compound-specific acceptable intakes. https://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM464285.pdf (accessed 2 October 2017).

12.

Calabrese

. On the origins of the linear no-threshold (LNT) dogma by means of untruths, artful dodges and blind faith. Environ Res 2015; 142: 432–442.

13.

Calabrese

EJ.

LNTgate: how scientific misconduct by the U.S. NAS led to governments adopting LNT for cancer risk assessment. Environ Res 2016; 148: 535–546.

14.

Boobis

. personal communication.

15.

Cheeseman

Machuga

Bailey

. A tiered approach to threshold of regulation. Food Chem Toxicol 1999; 37(4): 387–412.

16.

Bassan

Fioravanzo

Pavan

. SCIENTIFIC REPORT submitted to EFSA on applicability of physicochemical data, QSARs and read-across in threshold of toxicological concern assessment. http://www.efsa.europa.eu/fr/supporting/doc/159e.pdf.

17.

Fung

Barrett

Huff

. The carcinogenesis bioassay in perspective: application in identifying human cancer hazards. Environ Health Perspect 1995; 103(7–8): 680–683.

18.

Boobis

Brown

Cronin

MTD

. Origin of the TTC values for compounds that are genotoxic and/or carcinogenic and an approach for their re-evaluation. Crit Rev Toxicol 2017; 16: 1–23. DOI: 10.1080/10408444.2017.1318822

19.

EFSA CEF Panel (EFSA Panel on Food Contact Materials, Enzymes, Flavourings and Processing Aids). Identification of the substances from the Carcinogenic Potency Database (CPDB) which are of particular concern even if ingested at doses below 0.0025 µg/kg body weight per day. EFSA J 2016; 14(3): 4407. DOI: 10.2903/j.efsa.2016.4407/epdf.

20.

CPDB. Acrolein: https://toxnet.nlm.nih.gov/cpdb/chempages/ACROLEIN.html.

21.

Gold

Slone

Stern

. Comparison of target organs of carcinogenicity for mutagenic and non-mutagenic chemicals. Mutat Res 1993; 286(1): 75–100, https://toxnet.nlm.nih.gov/cpdb/pdfs/MutatRes93.pdf (accessed 2 October 2017).

22.

Snodin

Teasdale

. Mutagenic alkyl-sulfonate impurities in sulfonic acid salts: reviewing the evidence and challenging regulatory perceptions. Org Process Res Dev 2015; 19(11): 1465–1485.

23.

Benigni

Bossaa

Jeliazkova

. The Benigni/Bossa rulebase for mutagenicity and carcinogenicity – a module of Toxtree. EUR 23241 EN – 2008, https://eurl-ecvam.jrc.ec.europa.eu/laboratories-research/predictive_toxicology/doc/EUR_23241_EN.pdf (accessed 2 October 2017).

24.

EFSA Panel on Food Contact Materials, Enzymes, Flavourings and Processing Aids (CEF). Scientific Opinion on Flavouring Group Evaluation 63, Revision 2 (FGE.63Rev2): consideration of aliphatic secondary alcohols, ketones and related esters evaluated by JECFA (59th and 69th meetings) structurally related to saturated and unsaturated aliphatic secondary alcohols, ketones and esters of secondary alcohols and saturated linear or branched-chain carboxylic acids evaluated by EFSA in FGE.07Rev4. EFSA J 2013; 11(4): 3188. DOI: 10.2903/j.efsa.2013.3188/epdf.

25.

Amberg

Beilke

Bercu

. Principles and procedures for implementation of ICH M7 recommended (Q)SAR analyses. Regul Toxicol Pharmacol 2016; 77: 13–24, http://ac.els-cdn.com/S0273230016300277/1-s2.0-S0273230016300277-main.pdf?_tid=f7e1ed02-769f-11e7-9592-00000aacb361&acdnat=1501581689_32650fee2205bd471c7ecc68b08ba341 (accessed 2 October 2017).

26.

Ford

Bryan

Gregory

. Comparative evaluation of 11 in silico models for the prediction of small molecule mutagenicity: role of steric hindrance and electron-withdrawing groups. Toxicol Mech Methods 2017; 27(1): 24–35.

27.

Fraunhofer Institute. Time extrapolation factors in repeated-dose studies. http://www.item.fraunhofer.de/content/dam/item/en/documents/Annual%20Reports/JB_2011/JB_2011_152ff_time_extrapolation.pdf (accessed 2 October 2017).

28.

The Control of Substances Hazardous to Health Regulations 2002 (as amended); Approved Code of Practice and guidance. Health and Safety Executive, 2013. http://www.hse.gov.uk/pUbns/priced/l5.pdf (accessed 2 October 2017)

29.

Committee on Mutagenicity: thresholds for in-vivo mutagens (2010). https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/315698/assessment_of_threshold_for_in_vivo_mutagens.pdf (accessed 2 October 2017).

30.

Cao

. Genotoxicity of acrylamide and glycidamide: a review of the studies by HPRT gene and TK gene mutation assays. Gene Environ 2012; 34(1): 1–8.

31.

EFSA CONTAM Panel (EFSA Panel on Contaminants in the Food Chain), 2015. Scientific Opinion on acrylamide in food. EFSA J 2015; 13(6): 4104. DOI: 10.2903/j.efsa.2015.4104/epdf.

32.

Lipworth

Sonderman

Tarone

. Review of epidemiologic studies of dietary acrylamide intake and the risk of cancer. Eur J Cancer Prev 2012; 21(4): 375–386.

33.

Lipworth

Sonderman

Tarone

. Acrylamide: a human cancer risk? Eur J Cancer Prev 2013; 22(2): 193–194.

34.

Pelucchi

Bosetti

Galeone

. Dietary acrylamide and cancer risk: an updated meta-analysis. Int J Cancer 2015; 136(12): 2912–2922. DOI: 10.1002/ijc.29339/full.

35.

Pelucchi

Galeone

Negri

. Dietary acrylamide and the risk of endometrial cancer: an Italian case-control. Nutr Cancer 2016; 68(2): 187–192.

36.

Pelucchi

Rosato

Bracci

. Dietary acrylamide and the risk of pancreatic cancer in the international pancreatic cancer case-control consortium (PanC4). Ann Oncol 2017; 28(2): 408–414.

37.

CPDB. Acrylamide. https://toxnet.nlm.nih.gov/cpdb/chempages/ACRYLAMIDE.html (accessed 2 October 2017).

38.

European Medicines Agency. Conclusions and recommendations on the use of genetically modified animal models for carcinogenicity assessment, CPMP/SWP/2592/02/Rev 1, 2004. http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2009/09/WC500003257.pdf (accessed 2 October 2017).

39.

SCCP/SCHER/SCENIHR. Risk assessment methodologies and approaches for genotoxic and carcinogenic substances. 2009. http://ec.europa.eu/health/ph_risk/committees/04_scher/docs/scher_o_113.pdf (accessed 2 October 2017).

40.

Galloway

. International regulatory requirements for genotoxicity testing for pharmaceuticals used in human medicine, and their impurities and metabolites. Environ Mol Mutagen 2017; 58(5): 296–324. http://onlinelibrary.wiley.com/doi/10.1002/em.22077/epdf

41.

Dybing

Sanner

oelfzema

. T25: a simplified carcinogenic potency index: description of the system and study of correlations between carcinogenic potency and species/site specificity and mutagenicity. Pharmacol Toxicol 1997; 80(6): 272–279.

42.

Sanner

Dybing

Willems

. A simple method for quantitative risk assessment of non-threshold carcinogens based on the dose descriptor T25. Pharmacol Toxicol 2001; 88: 331–341.

43.

ECETOC (European Centre for Ecotoxicology and Toxicology of Chemicals). The use of T25 estimates and alternative methods in the regulatory risk assessment of non-threshold carcinogens in the European Union, Technical Report No. 83 (2002).

44.

Giordani

Kobel

Gally

. Overall impact of the regulatory requirements for genotoxic impurities on the drug development process. Eur J Pharm Sci 2011; 43(1–2): 1–15.

45.

Pirie

Peto

Reeves

. The 21st century hazards of smoking and benefits of stopping: a prospective study of one million women in the UK. Lancet 2013; 381(9861): 133–141.

46.

European Medicines Agency. Guideline on setting health based exposure limits for use in risk identification in the manufacture of different medicinal products in shared facilities. EMA/CHMP/ CVMP/ SWP/169430/2012. 2014. http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2014/11/WC500177735.pdf (accessed 2 October 2017).

47.

Draft guideline on assessment and control of DNA reactive, mutagenic impurities in veterinary medicines. EMA/CVMP/SWP/377245/2016. http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2017/02/WC500222249.pdf (accessed 2 October 2017).

48.

Scientific opinion on genotoxicity testing strategies applicable to food and feed safety assessment. EFSA J 2011; 9(9): 2379. http://www.sftox.com/actualites/EFSA_Genotox.pdf (accessed 2 October 2017).

49.

Hendriks

Derr

Misovic

. The extended toxtracker assay discriminates between induction of DNA damage, oxidative stress, and protein misfolding. Toxicol Sci 2016; 150(1): 190–203. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5009621/pdf/kfv323.pdf (accessed 2 October 2017)

50.

Hernández

van Steeg

Luijten

. Mechanisms of non-genotoxic carcinogens and importance of a weight of evidence approach. Mutat Res 2009; 682(2–3): 94–109.

51.

Greim

Hartwig

Reuter

. Chemically induced pheochromocytomas in rats: mechanisms and relevance for human risk assessment. Crit Rev Toxicol 2009; 39(8): 695–718.

52.

Committee on Hazardous Substances. Guide for the quantification of substance-specific exposure-risk relationships and risk concentrations after exposure to carcinogenic hazardous substances at the workplace. (2013): https://www.baua.de/EN/Topics/Work-design/Hazardous-substances/pdf/TRGS-910-Annex3.pdf?__blob=publicationFile&v=2 (accessed 2 October 2017).

53.

Müller

. Annual genotoxic impurities summit. Vienna, 1-2 June 2017.

54.

CPDB. 2,6-Dinitrotoluene: https://toxnet.nlm.nih.gov/cpdb/chempages/2%2C6-DINITROTOLUENE.html (accessed 2 October 2017).

55.

CPDB. 3,3′-Dimethylbenzidine.2HCl: https://toxnet.nlm.nih.gov/cpdb/chempages/3%2C3’-DIMETHYLBENZIDINE.2HCl.html (accessed 2 October 2017).

56.

CPDB. 3,3′-Dimethoxybenzidine.2HCl: https://toxnet.nlm.nih.gov/cpdb/chempages/3%2C3’-DIMETHOXYBENZIDINE.2HCl.html (accessed 2 October 2017).

57.

CPDB. 1,2-Dimethylhydrazine.2HCl: https://toxnet.nlm.nih.gov/cpdb/chempages/1%2C2-DIMETHYLHYDRAZINE.2HCl.html (accessed 2 October 2017).

58.

CPDB. 2-Hydroxyethylhydrazine: https://toxnet.nlm.nih.gov/cpdb/chempages/2-HYDROXYETHYLHYDRAZINE.html (accessed 2 October 2017).

59.

CPDB. Hydrazine: https://toxnet.nlm.nih.gov/cpdb/chempages/HYDRAZINE.html (accessed 2 October 2017).

60.

CPDB. Chlorambucil: https://toxnet.nlm.nih.gov/cpdb/chempages/CHLORAMBUCIL.html (accessed 2 October 2017).

61.

CPDB. 1,2-Dibromo-3-chloropropane: https://toxnet.nlm.nih.gov/cpdb/chempages/1%2C2-DIBROMO-3-CHLOROPROPANE.html (accessed 2 October 2017).

62.

CPDB. Benzotrichloride: https://toxnet.nlm.nih.gov/cpdb/chempages/BENZOTRICHLORIDE.html (accessed 2 October 2017).

63.

CPDB. 1,2,3-Trichloropropane: https://toxnet.nlm.nih.gov/cpdb/chempages/1%2C2%2C3-TRICHLOROPROPANE.html (accessed 2 October 2017).

64.

CPDB. 2-(2,2-Dimethylhydrazino)-4-(5-nitro-2-furyl)thiazole: https://toxnet.nlm.nih.gov/cpdb/chempages/2-(2%2C2-DIMETHYLHYDRAZINO)-4-(5-NITRO-2-FURYL)THIAZ.html (accessed 2 October 2017).

65.

CPDB. 2-Amino-5-(5-nitro-2-furyl)-1,3,4-thiadiazole: https://toxnet.nlm.nih.gov/cpdb/chempages/2-AMINO-5-(5-NITRO-2-FURYL)-1%2C3%2C4-THIADIAZOLE.html (accessed 2 October 2017).

66.

CPDB. Ethylene imine: https://toxnet.nlm.nih.gov/cpdb/chempages/ETHYLENE%20IMINE.html (accessed 2 October 2017).

67.

CPDB. β-Propiolactone: https://toxnet.nlm.nih.gov/cpdb/chempages/beta-PROPIOLACTONE.html (accessed 2 October 2017).

The cancer threshold of toxicological concern (TTC) in relation to foodstuffs and pharmaceuticals: A potentially useful concept compromised by a dubious derivation

Abstract

Keywords

Introduction

Use of the cancer TTC in regulatory toxicology

Criticisms of the TTC derivation

Skewed dataset for determination of cancer TTC

Lowest statistically significant TD50 versus default harmonic-mean TD50

Evaluation of non-mutagenic carcinogens

Approaches for pharmaceutical impurities and LTL exposure

Proposals for re-assessment of the cancer TTC by Boobis et al. 18

Discussion and Conclusions

Footnotes

Declaration of Conflicting Interests

Funding

References

Lowest statistically significant TD₅₀ versus default harmonic-mean TD₅₀

Proposals for re-assessment of the cancer TTC by Boobis et al.¹⁸