It is important to be aware that the SAR evaluation procedures can only be as good as the databases and rule sets that underpin the SAR systems. It is known that there are complexities in the models for some compound classes, for example those relating to anilines and heteroaromatic amines. It may be advisable to treat these cases where expert elicitation becomes essential with the option to consider safety testing (Ames test).
Although in silico systems are comprehensive in terms of the compound classes covered, there are nevertheless examples of classes that are not covered and for which there is no closely related data in the underlying database. This point was made by Dobo et al. [3] in respect of heteroaromatic nitro compounds. Hence, it is important for the recipients of the SAR output to scrutinize the findings. If an impurity has no flags for mutagenicity, but is used in the process as an electrophile, then it would be prudent to seek expert judgment with respect to the strength of the underlying data set. In such cases and particularly if the synthetic route is likely to remain the same up to and beyond marketing authorization, further assessment, i.e. an Ames test, may be prudent as this is likely to be ultimately required as part of worker safety expectations.
Evaluation of mutagenic risk can also be augmented by data derived from within the public domain. Indeed, such data forms the basis of the ICH M7 addendum table [8], and the review of common chemicals conducted by Bercu et al. [13]. This topic is explored in detail in Chapter 7.
Data sources include:
Hazardous Substances Databank (HSDB),
Chemical Carcinogenesis Research Information System (CCRIS), and
Integrated Risk Information System (IRS).
These provide an excellent source of safety data for many common chemicals. Another related system is the Berkeley database. Indeed, as described in Chapter 7, it is often possible with common reagents to locate sufficient safety data to allow mutagenic risk to be assessed on a compound‐specific basis rather than simply applying the TTC. Until recently these, and a number of other references, were accessible via TOXNET, a searchable database provided by the US Library of Medicine. TOXNET provided access to a series of databases through a common portal. While TOXNET is no longer available, alternative search engines such as TOXPLANET are available. Additionally, Lhasa has now reproduced the Berkeley database, which is accessible via their website. As well as the original Berkeley database, which is no longer maintained, the Lhasa database includes more recent data available since the freezing of the Berkeley database. See Chapter 7 for a detailed overview.
3.2.5 Step 4 – Assessment of Risk of Potential Carryover of Impurities
Once impurities with a potential mutagenic safety concern have been identified by the SAR evaluation process, the next step is to consider the likelihood of them being present in the isolated API, often referred to as impurity fate mapping.
The impurities under consideration are often highly reactive, and hence their removal during downstream processing is facilitated by this intrinsic reactivity. This removal can also occur as a result of a variety of factors including solubility, through extractive processes, or within the isolation solvent, i.e. mother liquors during isolation of the desired product, volatility, etc. For example, acidic and/or basic workup conditions frequently encountered in manufacturing processes may lead to decomposition and/or removal of the material of concern. Similarly, other reagents used in downstream processing may react with the material rendering it nonmutagenic, and thus the resulting impurity can be controlled to levels aligned with ICH Q3A or the appropriate clinical phase. It is important as part of such an assessment to consider the fate and effects with respect to what the downstream product could be “reasonably predicted” to be. While rare, it is possible that a PMI could be converted to another PMI through processing, e.g. oxiranes ring opening with HCl to the chloro‐alcohol.
It is important that some consideration should be given to what the impurity might be converted to. Factors that contribute to removal of such impurities are reviewed below in the following section. A more detailed examination is provided in Chapter 9.
Initially, such impurity fate assessments were largely based on the theoretical knowledge and experience of the evaluating chemist. Unfortunately, however compelling the arguments developed, they were viewed as nonquantitative and subjective from a regulatory perspective. Thus, in many cases there is a need to provide further analytical data to substantiate the impurity fate assessment. Hence, a quality by testing (QbT) approach was adopted rather than a quality by design (QbD) approach.
It was against this context that Teasdale et al. [14, 15] looked to define a potentially standardized approach to such assessments. The aim was to assess fate semiquantitatively based on factors linked to the impurity's physicochemical properties (and taking into account those of the API and intermediates) and the process conditions employed in the route of manufacture to the API. Pierson et al. [4] had earlier suggested that an assumption could be made of a 10‐fold reduction per synthetic stage. In many cases this would suffice and indeed may even be a cautious estimate of the risk. However, in certain circumstances, for example an unreactive mutagenic reagent or intermediate used in a “telescoped” process (no isolations between stages), this may be too simplistic and may even overestimate the potential purge. For this reason, a more quantitative approach, based on actual process conditions and the physicochemical properties of the MI in question, was sought and is outlined below.
A number of contributory factors have been defined that should be taken into account for such an assessment; these are described in detail in Chapter 9.
3.2.6 Overall Quantification of Risk
As described above the acceptability of chemistry‐based arguments to demonstrate purge of MIs was initially only partially successful due to its empirical nature. In order to make a quantitative assessment of the level of carryover of a particular material into an API or downstream intermediate, Teasdale et al. [14, 15] defined a number of mitigating criteria; these are defined in Table 3.4. This scoring system has been widely used [16–18], and the concept is enshrined within ICH M7, aligning with control Options 3 and 4; see Chapter 2 (ICH M7) and Chapter 9 (purge factor concept).
For each mitigating criteria, a purge factor can then be selected according to the characteristics of the material under consideration. The numerical scale has been developed to link individual process steps to the physicochemical properties of the individual impurity in question. Each