Читать книгу Mutagenic Impurities - Группа авторов - Страница 65

It is important to note that the guideline specifically stipulates that where adequate carcinogenicity data exist it should be used to calculate a compound‐specific AI or ADI. It also outlines that the approach should mirror that of the derivation of the TTC itself, i.e. linear extrapolation to a risk value of 1 in 100 000 analogous in risk terms to the TTC. It also states that other established riskvassessment practices such as those used by international regulatory bodies may be applied either to calculate AIs or the actual values themselves used. This apparently helpful statement does in fact lead to considerable confusion: should for example the value for a particular compound specified by the US environmental protection agency (EPA) (or other agencies) be simply adopted or in such instances should the available data be evaluated using the linear extrapolation? The guideline provides no clear statement on such a point and nor does it provide any specific example. In practice it would seem appropriate to use the default approach of linear extrapolation where data are available. Dobo et al. [27] recently reported on various ADIs that can be generated for hydrazine, exemplifying the considerable ambiguity that can be found when trying to generate a compound‐specific ADI for regulatory use.

Linked to this section is Note 4 where a specific example calculation is provided. The calculation outlines the determination of an AI for ethylene oxide. It is surprising that ethylene oxide was chosen as it is a gas, with good purging potential and of little synthetic utility, making its presence in final product very unlikely. Furthermore there is strong evidence that it is also generated endogenously [28]. In terms of the calculation itself, it is relatively straight forward. Terminal dose (TD₅₀) values are taken from the Carcinogenicity Potency Database (CPDB) for both rat and mouse, with the more conservative value being selected, 21.3 mg/kg/day (rat) and the limit calculated by dividing by 50 000 to adjust to a 1 in 100 000 risk and multiplied by the internationally accepted average human body weight (50 kg), to give an ADI of 21.3 μg/day for lifetime exposure.

On the face of it, this looks relatively straightforward; however, this is a simple example. In reality this is often far more complex. In many cases data are available for multiple carcinogenicity studies, within the CPDB these are combined and reported in terms of the harmonic mean. The studies involved may be of variable quality, e.g. insufficient duration, low animal numbers, and limited number of doses studied. Another important factor for consideration is tumor site and relevancy, for example forestomach tumors in rodents. Such tumors are often associated with local irritation/inflammation and are considered nonrelevant to humans, both from a physiological perspective (humans have no forestomach) and from an exposure perspective; the impurity at the low levels observed within a pharmaceutical product renders it extremely unlikely to result in such irritation.

As a result of the complexity described, a cross‐industry initiative was established that looked to develop an addendum table to the guideline. Included within this would be agreed limits for a range of common mutagenic/carcinogenic reagents. In addition to the actual agreed AIs, specific criteria were established to allow for the calculation of limits for other reagents in addition to those captured in the addendum. This ultimately culminated with the publication of the revised guideline in July 2017. The derivation of limits is described in detail in Chapter 7.