Читать книгу Mutagenic Impurities - Группа авторов - Страница 136

As shown within the previous case study (GW641597X), the normal process for a risk assessment would be to identify the potential impurities within the drug substance and subsequently establish where mutagenicity concerns exist. However, this can also be extended to identify component constituents that may react together to an impurity of concern. By applying ICH M7 [8] control principles to these reactive species, the MI risk and any necessary control strategies can be established.

Candesartan cilexetil was developed as an angiotensin‐II receptor antagonist for the treatment of hypertension. Following the discovery of N‐nitrosodimethylamine (NDMA) in batches of valsartan and subsequently in additional sartans (losartan and irbesartan), it became necessary for all sartan‐containing medications to evaluate the risk posed by nitrosamines, which form part of the cohort of concern, in drug products [24]. This case study evaluates the risk of nitrosamine formation and subsequent carryover to the API by examining the fate of the individual components required to generate a nitrosamine impurity.

Candesartan cilexetil is prescribed for chronic use and is therefore subject to lifetime TTCs for any impurities present within the drug product. However, regulatory guidance at the time also indicated that due to the potent mutagenic and carcinogenic potential of some nitrosamines, LTL limits for nitrosamines could not be used. Interim limits for the presence of nitrosamine impurities NDMA and N‐nitrosodiethylamine (NDEA) were set at 96 and 26.5 ng/day, respectively, based on extrapolation from the respective TD₅₀s. For candesartan, which has a maximum daily dose of 32 mg, this equates to a final API impurity concentration of 3 ppm for NDMA and 0.83 ppm for NDEA.

The most common method by which nitrosamines are formed is through the reaction of secondary or tertiary amines with a nitrosating agent such as sodium nitrite in acidic media. The process used for candesartan incorporates the use of triethylamine and dimethyl formamide (DMF), which are known to contain as contaminants, or decompose into, diethylamine and dimethylamine, respectively. The process additionally utilizes sodium nitrite, thereby introducing a theoretical risk of nitrosamine formation (Figure 3.7). The risk assessment process therefore needs to address whether these amines or the parent compounds are likely to be present within the same step (under appropriate conditions) at a level of concern, and thereby identify the degree of risk present for nitrosamine formation and any subsequent removal if formed.

Figure 3.7 Nitrosamine formation pathways from Et3N and DMF.

The manufacturing process for candesartan involves a nine‐stage synthesis whereby triethylamine and DMF are introduced in Stage 2, whereas nitrite is not present until Stage 5 (Figure 3.8). As a result, the key questions related to assessing the risk of nitrosamines within the synthesis of candesartan are:

1 Do Et3N, DMF, or their secondary amine degradants persist at an appreciable level into Stage 5?

2 Do traces of NaNO2 persist at an appreciable level through to Stage 7?

3 If yes to either Q1 or Q2, could a nitrosamine formed be expected to be present in the final API?

Figure 3.8 Process map of candesartan synthesis.

The use of purge assessments allows the adequate assessment of these questions.

Triethylamine is introduced into the synthesis at Stage 2 through its use as a base at stoichiometric quantities. However, following completion of the reaction, the crude product is obtained through concentration under full vacuum at 70–75 °C, which is able to remove the vast majority of the Et₃N (boiling point 89 °C). While no measured data was available, a conservative estimate based on the volume reduction indicated a level of 5% Et₃N for the crude product after concentration. This value (50 000 ppm) was therefore utilized as the starting concentration for the purpose of purge calculations (Figure 3.9 – step 2). The crude material then undergoes a range of unit processes to afford the clean intermediate 2 presenting a number of mechanisms of purge for Et₃N. An initial extraction with HCl_(aq) would result in formation of the corresponding salt (Et₃N.HCl), which is known to be highly soluble in water and can therefore reasonably be expected to purge to a high degree (Purge factor [PF] = 10) based on solubility as a result of its ionizability (Figure 3.9 – step 3). However, a subsequent basic extraction would not result in a similar purge, as the dominant free base shows excellent solubility in both organic and aqueous solvents making the distribution more even. A purge of 1 is therefore the highest value that can be assigned, despite the likely removal of some Et₃N in this process, to ensure a conservative prediction (Figure 3.9– step 4). Subsequent removal of the EtOAc (boiling point 77 °C) under reduced pressure will result in the co‐evaporation/azeotroping of triethylamine given the closeness of their respective boiling points. Utilizing the scoring system, a purge of 3 is scored for this step (Figure 3.9 – step 5), whereas the subsequent uptake and concentration in methanol (boiling point 66 °C) does not warrant application of purge as the boiling point of the triethylamine now exceeds 20 °C above that of the solvent (Figure 3.9 – step 6). Once again, this reiterates the conservative nature of the purge assignment, as some azeotroping is still likely to occur, particularly at reduced pressures where the difference in boiling points will contract to within 20 °C [14, 15]. Following precipitation and filtration of the intermediate, the Et₃N that is both highly soluble in methanol and a liquid itself can safely be considered to remain extensively within the mother liquors. Additionally, the subsequent wash of the filter cake to remove residual mother liquors and surface impurities allows for a further cautious score of 10 based on solubility (Figure 3.9 – steps 6 and 7).

Figure 3.9 Breakdown of purge assignments for Et₃N in the Stage 2 workup processes.

In the workup processes following Stages 3 and 4, the purge of Et₃N is observed through similar mechanisms, reliant on the high degree of solubility in the process solvents and low boiling point. The total predicted purge for triethylamine up to the point of introduction of NaNO₂ in Stage 5 is 8.1 × 10⁸ against a required purge of 60 240 to achieve the 0.83 ppm limit for NDEA within the API. Utilizing the approach to reporting for Option 4 strategies developed by Barber et al., this corresponds to a purge ratio of 13 446 for Et₃N (Figure 3.10). At this ratio very little justification would be necessary to demonstrate control of the impurity. In the scenario detailed here, the triethylamine is not the impurity of concern, but the nitrosamine NDEA that may be formed from it. Utilizing the same limit for the parent amine as for the nitrosamine imparts a further degree of conservatism, as quantitative conversion is hugely unlikely to occur, and therefore NDEA formation from Et₃N is demonstrated to be well controlled and suitably de‐risked.

Figure 3.10 Purge calculation summary for Et₃N.

A similar assessment was performed for both DMF (potential source of dimethylamine [DMA]) and the amines of concern, assuming their presence within the starting materials or from degradation (Figure 3.11). In each of these cases the degree of purge establishes the risk of carryover into Stage 5 to be low. In the case of DMF, which is approximated at a concentration of 200 000 ppm following the Stage 2 reaction, a target concentration of <1 ppm (below the 3 ppm limit for NDMA) and a predicted purge of 7.3 × 10⁹ equates to a purge ratio of 36 500. This demonstrates the potential for NDMA formation, resulting from DMA formed by the degradation of DMF in Stage 5, to be insignificant and requiring minimal justification.

Figure 3.11 Purge calculation summary for DMF.

The purge appraisal of DMA and DEA highlights their greater propensity to be removed, primarily linked to their low boiling points (Figure 3.12). Determining a ratio for these impurities is difficult, as a starting concentration cannot be determined; however, they cannot be present in greater quantities than their parent structures and yet the potential for purge is far greater. As such, any purge ratio derived would be far in excess of those obtained for Et₃N and DMF and therefore posing no appreciable risk to nitrosamine formation.

Figure 3.12 Purge calculation summary for DMA and DEA.

Purge calculations of amine‐related impurities within this synthesis has clearly demonstrated there to be no risk of formation of NDMA or NDEA within Stage 5, as the initial question has been answered – amine impurities and sodium nitrite are not present together within the same stage.

In order to fully de‐risk the formation of nitrosamines in the API, the formation of nitrosamines must also be considered within Stage 7, where both Et₃N and DMF are reintroduced into the synthesis. Once again this can be assessed by considering the ability for carryover of one of the reacting components, in this case the NaNO₂. The purge assessment of NaNO₂ (Figure 3.13) indicated a high degree of purge in the two steps, with a predicted purge of 1 × 10⁶. While the purge ratio for nitrite at this point is only 1, this does not preclude an Option 4‐type approach, merely necessitating a greater degree of evidence to support the assessment (see Barber et al. [19]). Purge assessments are conservative due to the limitation in purge value that can be assigned at each step, whereas in practice the true purge may be far greater. If this can be demonstrated, then the application of an Option 4 approach remains valid. To this effect, nitrite testing during Stage 6 found it was not present above 100 ppm (limit of detection [LoD]), thereby confirming the conservatism within the Stage 5 assessment.

Figure 3.13 Purge calculation summary for NaNO2.

The use of purge calculations therefore, in conjunction with analytical testing, established the formation of NDEA or NDMA cannot occur to a level of concern within the synthesis of candesartan, as there are sufficient levels of control of the component parts (amines and NaNO₂) to ensure that they are never present within the same stage at a concentration of concern, something that can be easily conveyed through a simple schematic (Figure 3.14).

Figure 3.14 Schematic of candesartan process highlighting the purge‐based risk assessment for nitrosamine formation and clearance.

The de‐risking process described for candesartan was further validated through trace analytical testing for NDMA and NDEA. While no risk of nitrosamine formation was identified within the candesartan synthesis, had the potential for formation been established, the purge principles could have been further exploited to determine the risk of carryover of the nitrosamines themselves into the final API, as any nitrosamine formed would still have the opportunity to be purged and controlled in subsequent stages. In the case of candesartan, a purge assessment of NDMA and NDEA from Stage 5 onward indicates theoretical purge factors of ~10 000 and ~1000, respectively.

In addition, analytical testing of over 100 batches of candesartan have confirmed the absence of NDMA or NDEA above 5 ppb (LoD), thereby validating the expert theoretical assessment that they could not be formed to a level of concern.