of potential MIs. For example, if an oncology treatment is itself known to be genotoxic, it would seem unnecessary to control potentially MIs in such an active substance to levels consistent with the TTC. Furthermore, many oncology treatments are used either post or in tandem with cytotoxics during the clinical phase, particularly in advanced stages of the disease. The cytotoxic agent itself poses a significant, but accepted risk, of secondary cancer. This again challenges the value to the patient, of controlling MIs to levels consistent with the TTC.
1.1.4.1.5 Control Requirements When Multiple GIs May Be Present
Given the complex multistep nature of the synthesis of many synthetic Active Pharmaceutical Ingredient (APIs), it is possible for a product to contain more than one potentially MI. Indeed, a study published in 2016 [11] showed that multiple reagents of mutagenic potential are used in a “typical” synthesis. The EMA guidance [9] was not clear on what control expectations would exist when more than one potential MI was likely to be present in the active substance on product. Would each be simply controlled on the basis of individual TTC limits?2
This would seem reasonable given the conservative nature of the derivation of the general 1.5 μg/day TTC limit. Or would there be an expectation that the total genotoxic impurity load would be controlled to a total level of 1.5 μg/day or other limit? There might be some scientific basis for implementing such a cumulative control if the impurities were known to be (or likely to be) toxicologically similar, but far less need to do so if the impurities were known to be (or likely to be) toxicologically distinct. These are all interesting and potentially important considerations, but the published guideline provided no detailed guidance on these questions. In terms of the toxicological risk, Bercu et al. reviewed the supporting evidence and demonstrated that with the addition of one to two MIs, a slight but insignificant increase in cancer risk was observed. There was not an increase in cancer risk when comparing structurally related impurities with structurally unrelated impurities. They therefore concluded that there was little evidence to support a view that effect was cumulative at low levels, <5 μg/day [12].
1.1.4.1.6 Application to New Marketing Authorisation Approval (MAA) Applications Relating to Existing Products
One of the specific challenges of the guideline was that it also potentially applied to applicants for generic versions of existing products. On one level, an applicant for a generic medicine might assume that the active substance in their medicine is “out of scope,” as clearly such a medicine has a significant preexisting period of use such that its safety is known. However, this assumption relies upon the generic active pharmaceutical ingredient and medicinal product having the same quality and impurity profile as the existing drug substance and drug product. This may on many occasions not be the case, as even if similar chemistry is used, subtleties of manufacture or formulation can lead to potentially significant differences in impurity profile, especially when “significant” is no longer being considered as reflecting levels commensurate with that of the ICH unspecified impurity control limits [2] (e.g. in the order of 0.1% – i.e. parts per thousand) but at the levels of TTC‐based controls (which can be in the order of parts per million and indeed for those impurities defined by the cohort of concern, parts per billion).
A particular challenge in terms of development of generic products relates to how can a generic applicant assure themselves they have introduced no new risk factors with respect to previously approved materials? Could this be achieved by simply meeting the preexisting European Pharmacopoeia (EP) monograph for the active substance (if one exists)? In reality even now it is likely that this will not be sufficient: monographs rarely include controls on potential MIs at low levels. Thus, many of these potential risks may be “invisible” in terms of the public quality standard; indeed, the recent issues surrounding N‐nitrosamines clearly highlight this. Maybe the generic applicant could simply test their drug substance against the previously approved drug substance? But what analytical methods should be used? Of course, this lack of transparency relates not only to the generic manufacturer, the regulator charged with assuring the suitability of the new product faces a similar challenge.
Of course, if the generic applicant decided to do a comprehensive and independent risk assessment of their drug substance or drug product and their manufacturing processes and establish TTC‐based controls for any potentially MIs (on the basis of structural alerts, etc.), then no doubt the regulatory agencies will be presented with a potentially approvable drug substance, associated specification, and manufacturing process. What even now remains unclear is how will the agency view the previously approved marketing application holders. Issues surrounding valsartan and N‐nitrosamines show that certainly where the risk is deemed to be a general risk this will very likely lead to a request to test all current approved products for the MIs that the subsequent applicant has determined to be potentially present.
1.1.4.2 Control Expectations for Excipients
When finalized the EMA guideline stated that it did not apply to excipients used in pharmaceutical manufacture, this being addressed by a separate EMA publication [13] (discussed further below). Clearly, some excipients are also manufactured by chemical synthesis and may therefore also be exposed to routes of manufacture that contain reactive and “at risk” reagents and intermediates. Global pharmacopoeias such as the EP and United States Pharmacopeia (USP) contain many synthetic excipients; some like polyethylene glycol (PEG) are polymers of epoxides or use epoxides to derivatize other materials (e.g. cyclodextrins). Epoxides are alkylating materials and hence are potentially mutagenic potential impurities in the excipients. Clearly with excipients often being a more significant percentage in weight terms of a medicine than the active substance, the potential risk associated with excipient impurities might also be of concern. Despite these potential risks, it is clear that many of the excipients have a significant history of safe use, many indeed are listed within the Food and Drug Administration (FDA) generally regarded as safe (GRAS) list [14] and hence the guidance ultimately concluded that there are no significant issues associated with well‐precedented excipients given their long‐established and demonstrated safety profile.
But what of the potential risk associated with manufacturing process changes related to excipient manufacture? What of the increasing case of a novel excipient being developed? In the case of a novel excipient, it is very likely that the expectations for the assessment of MIs should mirror that of any new chemical entity (NCE).
1.1.4.3 Control Expectations for Natural/Herbal Products
Pharmaceuticals are in the majority of cases well‐characterized small molecules manufactured by well‐defined chemical synthesis. However, the situation can be quite different in relation to some medicines derived from natural sources. Some of these natural product‐derived medicines, including herbal medicines, can be less well‐characterized materials and are often complex mixtures that may vary batch to batch in terms of components. Of course, the control of impurities in such medicines is also important and by extension one perhaps should consider whether such medicines too might contain potentially MIs [15]. It is, however, practically impossible to apply the same degree of risk management to the manufacture/isolation of a natural product, nor the same degree of process selection and design. How should one approach the management of potential mutagenic risk in such active substances? The EMA guideline [9] provides no specific guidance and therefore EMA published a separate guideline for such products.
1.1.4.4 Identification of Potential Impurities
As highlighted earlier the guideline noted that risk assessment of manufacturing processes should be undertaken to identify potential MIs
and that impurity structures should be risk assessed (using in silico systems that link structural motifs to potential toxicological responses). This sounds very reasonable and practicable. However, one could find two “experts” in the field who might draw up two different