Journal Pre-proof

Control of N-nitrosamine impurities in drug products: Progressing the current CPCA

framework and supporting the derivation of robust compound specific Acceptable
Intakes

David J. Ponting, Andreas Czich, Susan P. Felter, Susanne Glowienke, James S.

Harvey, Raphael Nudelman, Joerg Schlingemann, Stephanie Simon, Graham F.
Smith, Andrew Teasdale, Robert Thomas
PII: S0273-2300(24)00203-4
DOI: https://doi.org/10.1016/j.yrtph.2024.105762
Reference: YRTPH 105762

To appear in: Regulatory Toxicology and Pharmacology

Received Date: 17 June 2024

Revised Date: 27 November 2024
Accepted Date: 7 December 2024

Please cite this article as: Ponting, D.J, Czich, A., Felter, S.P, Glowienke, S., Harvey, J.S, Nudelman,
R., Schlingemann, J., Simon, S., Smith, G.F, Teasdale, A., Thomas, R., Control of N-nitrosamine
impurities in drug products: Progressing the current CPCA framework and supporting the derivation
of robust compound specific Acceptable Intakes, Regulatory Toxicology and Pharmacology, https://
doi.org/10.1016/j.yrtph.2024.105762.

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.

© 2024 Published by Elsevier Inc.

 Business Use

1 Control of N-nitrosamine impurities in

2 drug products: Progressing the current
3 CPCA framework and supporting the
4 derivation of robust compound specific
5 Acceptable Intakes
6

7 David J Ponting*1, Andreas Czich2, Susan P Felter3, Susanne Glowienke4, James S Harvey5, Raphael

f
8 Nudelman6, Joerg Schlingemann7, Stephanie Simon7, Graham F Smith8, Andrew Teasdale9, Robert

oo
9 Thomas1

10 * Corresponding author: [email protected]

r
11 Affiliations
-p
12 1 Lhasa Limited, Granary Wharf House, 2 Canal Wharf, Leeds, UK
re
13 2 Sanofi Deutschland GmbH, R&D Preclinical Safety, 65926 Frankfurt, Germany
lP

14 3 Procter and Gamble, Central Product Safety, 8700 Mason-Montgomery Rd, Mason, OH, USA

15 4 Novartis Pharma AG, Lichtstrasse 35, CH-4056 Basel, Switzerland

na

16 5 GSK R&D, Gunnels Wood Road, Stevenage, Herts, SG1 2NY, UK

ur

17 6 Nudelman ChemTox Consulting, Rehovot 7628422, Israel

18 7 Merck Healthcare KGaA, Frankfurter Straße 250, 64293 Darmstadt, Germany

Jo

19 8 AstraZeneca, Safety Innovations, Clinical Pharmacology and Safety Sciences, R&D, Cambridge,
20 United Kingdom

21 9 Chemical Development, Pharmaceutical Technology & Development, Operations, AstraZeneca,

22 Macclesfield, United Kingdom

23 ORCIDs
24 DJP: 0000-0001-6840-2629

25 AC: 0000-0001-5106-3325

26 SPF: 0000-0001-6431-2028

27 SG: 0009-0006-3594-1443

28 JSH: 0000-0002-6656-375X

29 RN: 0000-0002-6005-7725

30 JS: 0000-0003-4799-1086

1
 Business Use

31 SSi: 0000-0002-2497-5211

32 GFS: 0000-0003-0393-3650

33 AT: 0000-0003-0309-8712

34 RT: 0000-0003-0261-8720

35 CRediT
36 DJP: Conceptualization, writing – original draft, writing – review and editing

37 AC: writing – review and editing

38 SPF: Conceptualization, writing – original draft, writing – review and editing

39 SG: writing – review and editing

40 JSH: Conceptualization, writing – review and editing

f
oo
41 RN: Conceptualization, writing – review and editing

42 JS: Visualization, Writing – Original Draft, Writing – Review & Editing

r
43 SSi: writing – review and editing -p
re
44 GFS: Conceptualization, review and editing

45 AT: writing – review and editing

lP

46 RT: Visualization, Writing – Review & Editing

na

47

48 Highlights
ur

49 - The CPCA provides a tiered TTC-like triage for nitrosamine impurity assessment.
Jo

50 - de minimis risk TTC-like limits should not be definitive safety thresholds.

51 - Risk assessment is an iterative process – CPCA refinements are possible.

52 - Information from close analogues is an important part of risk assessment.

53 - Compound-specific data, from in silico to in vivo studies, refines initial triage.

54

55

56 Abstract
57 The carcinogenic potency categorisation approach (CPCA) has recently been introduced by health

58 authorities. In this model, structural features from recent literature, industry proposals, and analyses

59 performed by health authorities, provide a rapid assessment of the potential acceptable intake (AI)

60 for a nitrosamine impurity. As with other screening regulatory values (such as the ICH M7 Threshold

2
 Business Use

61 of Toxicological Concern), the CPCA is conservative and can be considered a de minimis risk

62 management framework. In cases where a nitrosamine drug substance-related impurity (NDSRI) is

63 present below the CPCA limit, the framework provides resolution from a toxicological perspective (i.e.,

64 no further toxicology studies are required). Where an NDSRI is above the CPCA limit, the framework

65 provides for the initiation of additional activities (i.e., the CPCA is not the only possible limit). Read-

66 across approaches are described in both the CPCA and M7 guidance and can provide a limit with more

67 specific applicability than the general model. The use of available experimental data (in vitro or in

68 vivo), is valuable in order to provide an even more specific limit. The CPCA provides a framework;

f
oo
69 however, data should permit changing the AI from initial structural assessment, based on increasing

70 data, to ultimately increase precision of the AI.

r
71 Introduction
-p
re
72 The carcinogenic potency categorisation approach (CPCA)1 for the assessment of nitrosamine
lP

73 (NA) impurities was a pivotal event that provided significant resolution to five years2 of uncertainty

74 for the pharmaceutical industry, whereby up to 40 % of drugs3 were potentially within scope of
na

75 significant Chemistry Manufacturing and Control remediation activities that could have resulted in
ur

76 concomitant shortages of whole classes of drugs (e.g., the ACE-inhibitor ‘pril’ drugs (quinapril,
Jo

77 ramipril…), or the beta-blocker ‘olol’ drugs (propranolol, atenolol…)). While the presence of the N-

78 nitrosamine impurities in these drugs was known, and additional toxicological assays were being

79 performed,4 no official announcement of appropriate lifetime Acceptable Intakes (AIs) had been made

80 for these impurities, implying a class-specific lifetime TTC limit of 18 ng/day5,6 might eventually need

81 to be applied. The announcement and adoption of the CPCA framework by European Medicines

82 Agency (EMA) was rapidly followed by additional Health Authorities (HAs) including Health Canada

83 (HC) and the U.S. Food and Drug Administration (FDA). Indeed, it was quickly made clear that all major

84 authorities had been involved in its development via the global Nitrosamines Implementation

85 Technical Working Group (NITWG).1 Stakeholders considered significant progress had been made

86 regarding the assessment and control of the NDSRIs mentioned in the beta-blockers and ACE inhibitors

3
 Business Use

87 as the AIs for these NA impurities were clarified and re-orientated to a limit comparable to the general

88 threshold of toxicological concern (TTC) for mutagenic impurities, 1500 ng/day, significantly reducing

89 the uncertainty regarding the potential for class-wide market withdrawals for these drugs.7 Thus, the

90 introduction of the CPCA represents a significant advancement in the assessment of NA impurities by

91 defining regulatory defaults that can be used by HAs and MAHs to readily derive conservative AIs that

92 are likely to find regulatory endorsement.

93 Managing drugs where the CPCA AI limits for N-nitrosamine impurities remain at 400 ng per

94 day, or below 6 can be problematic; not only because of difficulties associated with developing trace

f
oo
95 analytical methods7 but also from reducing levels to below the AI, given the ubiquity of nitrite in

r
96 excipients as well as from other sources including polluted air.8,9 In order to provide for continued

97
-p
supply of these medicines, including those deemed essential by the WHO3,7, alternative methods for
re
98 determining an AI are required. The regulatory guidance that introduced the CPCA provided both an
lP

99 acknowledgement that the CPCA is conservative,1 and would be subject to change as new data
na

100 becomes available and the science advances and also allowed for the use of alternate methods such

101 as read-across to determine AIs. In this commentary, while acknowledging the importance of the
ur

102 highly conservative CPCA framework, the authors make the case for future modifications to the CPCA
Jo

103 as well as the continued use of compound-specific toxicological data and read-across to surrogates

104 with robust data – where such surrogates exist - to be routine rather than the exception.

105 Read-across, as a process, can be separated into two key stages. Firstly, the selection of a

106 suitable surrogate, in terms of chemical similarity. This needs to address measures of similarity such

107 as the local environment and reactivity, physicochemical properties, and more2. Secondly, the data

108 quality on the proposed surrogate should be evaluated. For nitrosamines, few of the compounds that

109 have been tested fully meet the criteria of OECD TG451, many having been tested before it was

110 introduced in 1981, including compounds that have been used for read-across by health authorities

111 (e.g. NNK which was tested only in one sex and with only 30 animals in the top dose group of three,

4
 Business Use

112 or N-nitroso piperidine, where while there were 15 dose groups plus control these were all small, with

113 12 animals of mixed sex/group, a total count lower than the 300 plus control animals required

114 according to TG451). Indeed, strict application of the guidance would indicate that only the 4080-rat

115 study of Peto et al10 on NDEA and NDMA would be a suitable study according to this measure. This

116 collection of not-to-modern-standards data is however the best and only data that is available on the

117 carcinogenic potency of nitrosamines, and while not perfect does contain salient and useful

118 information for potency assessment. This has therefore been the data which has been used for the

119 derivation of the TTC approach and the cohort of concern11–13, the codification of that cohort into ICH

f
oo
120 M714, the 18 ng/day class-specific limit5,6,15, the SAR work that has been performed on nitrosamines16–

20
121 and ultimately is the data on which the CPCA itself is built1,15,21,22. A decision on whether a study is

r
122
-p
fit-for-purpose as an analogue for read-across is therefore a matter of expert judgement, considering
re
123 what information the study provides. If the initially-selected analogue has no suitable toxicity studies,
lP

124 the next-most-similar analogue should be evaluated in the same way, which can be repeated until

125 there are no more similar analogues available.

na
ur
Jo

5
 Business Use

f
r oo
-p
re
126
lP

127 Figure 1: Nitrosamines assigned to various categories by health authorities. FDA, U.S. Food and Drug
128 Administration. EMA, European Medicines Agency. Health Canada. HSA, Health Sciences Authority
na

129 [Singapore]. TGA, Therapeutic Goods Agency [Australia]. COFEPRIS, Comisión Federal para la
130 Protección contra Riesgos Sanitarios [Mexico]. MFDS, Ministry of Food and Drug Safety [South Korea].
131 ANVISA, Agência Nacional de Vigilância Sanitária [Brazil]. SAHPRA, South African Health Products
ur

132 Regulatory Authority. PMDA, Pharmaceuticals and Medical Devices Agency [Japan]. BPOM, Badan
133 Pengawas Obat dan Makanan [Indonesia]. NMI, non-mutagenic impurity. EAT, enhanced Ames test.
Jo

134 Since the publication of the CPCA, lists of AIs have been published by different regulators, the

135 majority of which are CPCA-assigned limits (Figure 1). However, the publication of these CPCA AI limits

136 alongside the established ICH M7 approaches (e.g., AI limits based on read-across to surrogates) has

137 also inadvertently re-introduced the challenge associated with using the ‘as low as reasonably

138 practical’ or ‘ALARP’ principle with respect to the control of NA impurities in drug products. In practice,

139 an NA impurity can have two AI limits, given equal weights in guidance, one based on a specific CPCA

140 categorisation, and another based on read-across to a structurally-similar surrogate with robust

141 carcinogenicity data, both of which would be considered suitably protective (i.e., commensurate with

142 a < 1 in 100,000 theoretical excess lifetime cancer risk). A determination needs to be made in these

143 cases as to which limit is the most scientifically relevant, which should then be adopted; however, in

6
 Business Use

144 such cases there has been a concern that the lowest and most conservative AI limit may be adopted

145 (i.e. a return to the principle of setting an AI limit based on process capability rather than setting

146 compound-specific toxicology based AI limits), something that the ICH M7 guidance explicitly aimed

147 to prevent – technical feasibility should not be the denominator for an AI, being even less

148 toxicologically relevant than a de minimis risk TTC. However, encouragingly, recent changes to the

149 acceptable intakes for the sitagliptin impurity NTTP and varenicline impurity N-nitrosovarenicline have

150 been to use the CPCA rather than the (lower) read-across intake by read-across from N-

151 nitrosotetrahydropyridine15,16,21,22 – for N-nitrosovarenicline this ultimately leads to an AI close to one

f
oo
152 that has been proposed by read-across from a different compound, N-nitrosohexamethyleneimine16.

r
153

154
-p
Risk assessment is a tiered and evolutionary process
re
155 Exclusion of N-nitroso compounds from the original TTC risk management framework
156 A tiered TTC approach was published by Kroes et al.11 as a way to prioritize the need for further
lP

157 evaluation of substances present in the diet at low levels; this approach has since been adopted by a
na

158 number of regulatory agencies for broader application beyond substances in the diet. Based on

159 structural considerations, four different tiers were established for compounds without structural alerts
ur

160 for DNA reactivity, ranging from 18 ug/day to 1800 ug/day. For compounds with structural alerts or
Jo

161 data indicating potential mutagenicity, a lower TTC limit was established based on evaluation of rodent

162 carcinogenicity data (of varying quality, as discussed) on 730 carcinogens. Within that carcinogenicity

163 database, Kroes and colleagues identified five structural groups as potentially having a carcinogenic

164 potency that exceeded the TTC tier established for low level exposures genotoxic/mutagenic

165 substances (1.5 µg/day for pharmaceutical impurities); these substances were defined as being in a

166 “cohort of concern.” One of those structural groups was N-nitroso compounds. Importantly, however,

167 of the 105 N-nitroso compounds evaluated at the time, only 47 of them (45 %) had a greater potency

168 such that the TTC limit would not be considered protective (i.e., associated with an acceptable

169 theoretical excess lifetime cancer risk). This contrasts with aflatoxins, for which 100 % (5/5) were

7
 Business Use

170 determined to be more potent than the TTC limit11,23 (although it should be noted that the sample size

171 for the aflatoxin and alkyl-azoxy classes is small).

172 While the N-nitroso compounds evaluated by Kroes et al.11 were smaller molecular weight NAs with

173 rodent carcinogenicity data of varying quality, a range of exposure limits (Table 1) was identified that

174 extended two orders of magnitude above the TTC limit (i.e., these NAs were much less potent and for

175 whom the TTC limit of 1.5 µg/day would still be quite conservative for many of them); this is consistent

176 with the broader range of rodent carcinogenic potency that has been described as extending over four

177 orders of magnitude for NAs5,24, with the other two orders of magnitude extending below the TTC limit

f
oo
178 of 1.5 µg/day. The exposure limits for 14% of those NAs would still be considered protective at a limit

r
179 an order of magnitude higher than the TTC (i.e., 15 µg/day for a pharmaceutical impurity) and very few

180
-p
NAs were identified with a potency two orders of magnitude lower (i.e., where an exposure limit of
re
181 150 µg/day would be protective). At the time, no investigation was done of the distribution of
lP

182 potencies of those NAs for which the TTC limit would not be protective instead a conservative decision
na

183 was made to simply designate the entire class as being in the cohort of concern and thus outside of

184 the scope of the TTC based risk management framework (i.e. that a chemical-specific assessment
ur

185 would be needed to establish an acceptable exposure limit for members of the chemical class). If a
Jo

186 specific NA could be determined to have a carcinogenic potency that is less than that associated with

187 the TTC limit for genotoxic/mutagenic impurities, then it follows that that the compound in question

188 could be considered not to be in the cohort of concern.

189 Table 1 has been modified from Table 1 of Kroes et al11. to show a subset of the structural groups

190 defined in that publication and to modify the limits to reflect a theoretical excess lifetime cancer risk

191 of 1 in 100,000 vs 1 in 1,000,000 in accordance with the ICH M7 guidance on the use of TTC for

192 genotoxic impurities in pharmaceuticals. The first three structural groups in Table 1 are in the cohort

193 of concern. The fourth row (aromatic amines) is representative of a structural group which is not in

194 the cohort of concern. In the case of aromatic amines, the TTC limit is protective for 97% (148/153) of

8
 Business Use

195 the rodent carcinogens in that class. While this table highlights the greater potency of rodent

196 carcinogens in the cohort of concern, it also emphasizes the wide range of potencies for N-nitroso

197 compounds such that more than half (58/105) are less potent than the TTC of 1.5 μg/day for genotoxic

198 impurities in pharmaceuticals, raising a question as to whether some compounds in this class can be

199 excluded from the cohort of concern altogether17. This has recently been reviewed in greater detail

200 by Snodin23.

201 Table 1: Numbers and fractions of compounds in different structural groups that are estimated to give a
202 theoretical excess lifetime cancer risk greater than 1 in 100,000 at different intake levels (modified from Table 1
203 of Kroes et al.11). N = Number of compounds with an associated theoretical risk > 1:100,000 at the indicated

f
204 level, F = the fraction of the same.

oo
Exposure Limit Associated with theoretical 1 in 100,000 increase in lifetime cancer risk Total
1.5 µg 15 µg 30 µg 60 µg 150 µg ≥ 300 µg

r
N F N F N F N F N F N F
Aflatoxin-
like
5 1 -p 5
re
Azoxy 4 0.80 4 0.80 4 0.80 5 1 5
N-nitroso 47 0.45 90 0.86 96 0.91 99 0.94 102 0.97 105 1 105
Aromatic 5 0.03 51 0.31 71 0.44 82 0.51 106 0.65 153 1 153
lP

Amines
205
na

206
ur
Jo

207
208 Figure 2: Proportion of compounds in varying classes with TD50 values less than the figure shown (after Kroes et
209 al11)

210

9
 Business Use

211 Subsequent regulatory guidance on the assessment and control of NA’s

212 In 2017 the ICH M714 guidance for the assessment and control of mutagenic impurities in in drug

213 products recognised that the AIs for certain chemical classes (i.e., aflatoxin-like-, N-nitroso-, and

214 alkyl-azoxy compounds) would likely be significantly lower than the general TTC of 1.5 µg/day

215 considered applicable for mutagenic impurities of all other chemical classes. The guidance instead

216 recommended that if these compounds were found in drug products that a case-by-case approach

217 using e.g., carcinogenicity data from closely related structures, if available, should usually be

218 developed to justify acceptable intakes for pharmaceutical development and marketed products.

f
oo
219 In 2020 the EMA6 published an initial default class-specific lifetime TTC of 18 ng/day 5 for NAs lacking

220 sufficient toxicology data to support a compound-specific risk assessment (based either on chemical-

r
221 -p
specific data or read-across to a suitable analogue). The class-specific lifetime TTC value of 18 ng/day
re
222 was based on the identification of the lower 5th percentile of the reported rodent TD50 values for NAs
lP

223 in the Lhasa carcinogenicity potency database (LCDB)5,6. Although the underlying data and statistical

224 methods associated with this EMA limit were not published, similar values could be derived using
na

225 multiple statistical methods following an analysis of the available data for N-nitroso compounds (i.e.
ur

226 N-nitrosamines, but also N-nitrosoureas, nitrosamides etc) in the LCDB and/or the original Gold CPDB
Jo

227 database.5,25. No additional class-specific lifetime TTC limits for NAs were published by other

228 regulatory agencies.

229 This single-tier default class-specific lifetime TTC exposure limit for NAs was subsequently

230 updated to a multi-tiered approach called the Carcinogenic Potency Categorization Approach (CPCA),

231 which was published by the EMA, Health Canada, and the US FDA1. The CPCA is a binning approach in

232 which NAs can be assigned to one of five potency binsi based solely on consideration of the chemical

233 structure, based on an assessment of the number of α-hydrogen atoms and activating or deactivating

234 structural features present in the NA. These have proven useful for the development of in silico tools

i
While there are five CPCA categories, it is noted that CPCA 4 and CPCA 5 have been assigned the same limit of
1500 ng/day.

10
 Business Use

235 from the major QSAR vendors, such as Derek Nexus26 (comparable tools are available from all major

236 vendors, and it is assumed that a comparable validation has been performed) by confirming the scope

237 of different features.

238 While a crucial development, it is important that the CPCA be regarded as a tool to help prioritize

239 further research and/or risk evaluations as needed. Each CPCA potency category is associated with an

240 AI as follows Category 1 = 18/26.5 ng/day; Category 2 = 100ng/day; Category 3 = 400 ng/day; Category

241 4 =1500 ng/day; Category 5 = 1500 ng/day. The rationale for the AIs for each CPCA potency category

242 has been by reference to three specific NAs (NDEA, NDMA and NNK) for CPCA categories 1 and 2.

f
oo
243 These compounds are considered by agencies to have robust mutagenicity and carcinogenicity data

r
244 that support the calculation of a TD50 and an associated AI. It should however be noted that NDMA

245
-p
and NNK, despite being compounds that if no carcinogenicity data existed would be in category 1, are
re
246 used as the definition for category 2; however, there is no clear separation between the categories
lP

247 (vide infra) and therefore the fact that these compounds are in category 1 does not a priori mean they
na

248 are more potent than all category 2 compounds; the same would apply to other category boundaries.

249 Beyond that, a scaling factor of approximately 4 was applied to the initial AI for CPCA categories1,
ur

250 which gives the limit for category 3 (i.e., 26 ng/day x4 = 100 ng/day x4 = 400 ng/day x4 = 1600 ng/day)
Jo

251 with the last value being adjusted to align with the TTC of 1500 ng/day defined in the ICH M7

252 guidelines for the control of mutagenic impurities.

253 The conservative nature of the CPCA framework - Categories 1 & 2

254 When compared to AIs based on small-molecule TD50s from the LCDB, the CPCA categories 1 and 2

255 appear to have succeeded in producing conservative limits, however, it is clear that even in small-

256 molecule chemical space the range of potencies within a category spans several orders of magnitude,

257 stressing that most NAs within a category will not be anywhere close to the category limit in potency,

258 and most less potent than the limits of higher categories. Figure 3 shows the distribution of compound

259 AIs compared to their respective category limits with AIs calculated from summary harmonic mean

260 Lhasa TD50s where available, using data from the Lhasa Carcinogenicity Database (LCDB) version

11
 Business Use

261 2024.1 (a comparable figure is available as Figure 6 of Kruhlak et al1). For compounds where no Lhasa

262 TD50 was available Gold values were used; although this provides a larger dataset for comparison it

263 does risk the introduction of less reliable estimates. As can be seen in the figure no category 1

264 compounds fall below the 18 ng/day limit. In category 2 two compounds fall below the 100 ng/day

265 limit; the identity of and reasons for these are discussed in section S1 of the supporting information.

266

267 Significant variation can be seen in both categories with each category spanning three orders of

268 magnitude in potency. As both categories appear log-normally distributedii we can quantify the

f
oo
269 variation based on the standard deviation of the log(AI). Category 1 has a geometric mean (the un-

270 logged value of the mean log(AIs)) of 593 ng/day – i.e. above the category 3 limit - with a standard

r
271
-p
deviation of 431 %, while category 2 has a geometric mean value of 915 ng/day with a standard
re
272 deviation of 462 %. Given this wide variation in potencies, an expected value such as a median or mean
lP

273 is of limited use, instead a low percentile of the category limit is of primary concern for assessing de

274 minimis risk, but does not represent the actual potency of many of the compounds in the category.
na

275 The 18 ng/day and 100 ng/day limits represent approximately the 2nd and 10th percentiles of their
ur

276 respective categories based on the distributions fitted to log(AI) using previously-published methods5.
Jo

ii
p = 0.41 and 0.73 based on a Shapiro-Wilk test27 of log(AI) for categories 1 and 2 respectively.

12
 Business Use

277

f
r oo
278
-p
re
279 Figure 31: Potency of known category 1 and 2 nitrosamines. The nitrosamines in both categories are log-normally
280 distributed, with the category limits representing the 3rd and 10th percentiles for category 1 and 2 respectively.
lP

281 Two category 2 compounds have TD50s placing them below their limit, these are 2-
282 hydroxypropylmethylnitrosamine (75411-83-5) and nitroso-2,3-dihydroxypropyl-2-oxopropylamine (92177-50-
283 9). Potencies are based on Lhasa TD50s (LCDB version 2024.1) where available, where Lhasa TD50s are not
na

284 available Gold TD50s are used.

285
ur

286 Risk assessment is an iterative process

Jo

287 In considering the role of CPCA-derived AIs versus those based on chemical-specific data and/or read-

288 across, it is important to recognize that risk assessment is an iterative process. This was emphasized

289 in the National Academy of Science (NAS)’s seminal publication on “Science and Judgment in Risk

290 Assessment”28 which emphasized the need for an iterative approach to ensure the best use of limited

291 resources: This

292 “An iterative approach to risk assessment would start with relatively inexpensive screening

293 techniques and move to more resource-intensive levels of data gathering, model construction, and

294 model application as the particular situation warranted. To guard against the possibility of

295 underestimating risk, screening techniques must be constructed that err on the side of caution when

296 there is uncertainty.”

13
 Business Use

297 Indeed, the initial guidance provided by regulatory agencies for addressing novel nitrosamines

298 discovered in drug products included only a single default Acceptable Intake value (18 or 26.5 ng/day,

299 depending on the geographyiii) for nitrosamines lacking adequate safety data and for which read-

300 across was not supported. While the CPCA represents a major step forward in regulatory guidance for

301 addressing the safety of novel NAs compared to having only a single regulatory default for all

302 nitrosamines, it is important to recognize that this is still a default-based approach, akin to a tiered-

303 TTC for NAs. The five CPCA tiers cover a range of about two orders of magnitude (from 18 or 26.5

304 ng/dayiii to 1500 ng/day) while it is widely recognized that the potency of NAs ranges over more than

f
oo
305 four orders of magnitude5,24 – with the least potent NAs being fully two orders of magnitude less

306 potent than the TTC. Accordingly, while the CPCA approach now defines multiple potency bins, t it

r
307
-p
was intentionally designed to be conservative and is still based solely on certain aspects of the
re
308 chemical structure. The limits in the CPCA tiered approach should be used the same way that other
lP

309 TTC-based exposure limits are used – i.e., it should be concluded that while exposures below the

310 appropriate TTC limit do not unacceptably raise theoretical excess lifetime cancer risk and are
na

311 deprioritized for further evaluation, exposures above a given TTC-based exposure limit should not be
ur

312 automatically seen as raising a concern regarding patient safety; rather, the finding requires further
Jo

313 consideration (e.g., in vitro or in vivo toxicology studies, as denoted in nitrosamine guidance15,21,29,

314 and/or Chemistry Manufacturing and Control remediation activities). This concept has been clearly

315 articulated in guidance on the general use of the TTC based risk management framework in a number

316 of regulatory opinions 22, 23

317 The CPCA for NAs mirrors the situation with the tiered TTC approach for non-genotoxic compounds,

318 which also ranges two orders of magnitude from 18 µg/day (for organophosphates) to 1800 µg/day

319 (for chemicals assigned to Cramer Class I). Meanwhile, the underlying database of chemicals has a

320 potency range of over four orders of magnitude.11,30 Furthermore, each of the three historical Cramer

iii
EMA and Health Canada established a limit of 18 ng/day while FDA established a limit of 26.5 ng/day for
CPCA

14
 Business Use

321 Class noncancer TTC tiers (90 µg/day, 540 µg/day, and 1800 µg/day for the oral route of

322 administration) has been shown to be conservative,11,31 with the potency ranging over two orders of

323 magnitude for each individual tier such that this creates an overlap – i.e., a chemical assigned to the

324 most conservative tier (Cramer Class III) could well have a potency lower than many chemicals in

325 Cramer Class II or even Cramer Class Iiv. This is not unexpected and reflects the highly (and

326 intentionally) conservative approach to assigning chemicals to a potency bin based on a fairly

327 simplistic evaluation of the chemical structure along with worst-case assumptions regarding potency.

328 This approach often does not consider all aspects of a chemical, some of are expected to make it less

f
oo
329 potent and often by orders of magnitude.

r
330 Likewise for genotoxic impuritiesv in pharmaceuticals, the ICH M7 guidance14 emphasizes this point,

331
-p
stating that “[t]he use of a numerical cancer risk value (1 in 100,000) and its translation into risk-based
re
332 doses (TTC) is a highly hypothetical concept that should not be regarded as a realistic indication of the
lP

333 actual risk. Nevertheless, the TTC concept provides an estimate of safe exposures for any mutagenic
na

334 compound. However, exceeding the TTC is not necessarily associated with an increased cancer risk

335 given the conservative assumptions employed in the derivation of the TTC value.” The guidance14
ur

336 describes a number of factors that have gone into ensuring that the TTC is reliable, and the methods
Jo

337 used to derive it are sufficiently conservative such that even in the absence of data, one can be assured

338 that exposures below the TTC pose a negligible theoretical risk of carcinogenicity or other toxic effects.

339 These same conservative methods were assumed to be used to establish the default class-specific

340 lifetime TTC of 18 ng/day for NAs and similarly have also been used to establish the new CPCA-based

341 limits that allow for binning NAs according to chemical structure. Thus, the CPCA is an extension of

342 existing tiered TTC approaches, the exposure limits for each have been established by regulators to

343 be highly conservative. As described in the ICH M7 guidance, the availability of TTC-derived limits

iv
Note that for noncancer TTC tiers, the more potent chemicals are assigned to the highest Cramer Class (Class
III) while for the new CPCA tiers, the more potent chemicals are assigned to the lowest CPCA Category.
v
Those that do not fall into the Cohort of Concern

15
 Business Use

344 does not preclude the development of compound-specific assessments, and in fact, states that

345 “Compound-specific risk assessments to derive acceptable intakes should be applied instead of the

346 TTC-based acceptable intakes where sufficient carcinogenicity data exist.” This could be on the basis

347 of chemical-specific carcinogenicity data, read-across or “other established risk assessment practices

348 such as those used by international regulatory bodies.”

349 Impact of the CPCA

350 The CPCA is, in summary, a tier-0 TTC-like risk management framework, which has

351 dramatically reduced the need for in vitro and in vivo toxicity studies and provided clarity regarding

f
oo
352 the expectations for the control of NDSRIs in drug products. This applied in all cases where the NDSRI

353 can be controlled to levels less than the CPCA limits but is particularly apparent with category 4 and 5

r
354 -p
NDSRIs, where the control limits are commensurate with established process capabilities
re
355 implemented as part of the ICH M7 process.
lP

356

357 Proposed refinements to the CPCA

na

358 It is important to recognize that, while a critical part of regulatory guidance, the CPCA is
ur

359 expected to evolve as the science advances. The recent expansion of the HESI GTTC MGRA

Nitrosamines initiative to include a QSAR/QM workstream32 reflects the need for this research. This
Jo

360

361 is clearly articulated in regulatory guidance15,21,22

362 “It is recognised that the science is evolving in the prediction of mutagenic potential and

363 carcinogenic potency based on SAR concepts. Therefore, the predicted Carcinogenic Potency

364 Categorization Approach described in this document is a conservative approach that represents

365 the best available science at this time and is expected to be further refined and expanded as new

366 data become available. This may include refinement of the AI limits associated with predicted

367 carcinogenic potency categories and changes to the structural features and their associated

368 activating and deactivating feature scores.”

369

16
 Business Use

370 Electron-withdrawing groups

371 The fact that the CPCA does not contain an exhaustive list of the features that can/should be

372 considered in assigning a potency category is also explicitly recognized in the current description of

373 the deactivating features that lead to a lower potency bin. For example, two of these deactivating

374 features are described as:

375 • Electron-withdrawing group** bonded to α-carbon on only one side of N-nitroso group (cyclic or

376 acyclic) [Score of +1]

377 • Electron-withdrawing groups** bonded to α-carbons on both sides of N-nitroso group (cyclic or

f
378 acyclic) [Score of +2]

oo
379 Examples are provided, but a comprehensive list of what constitutes an “electron-

r
380
-p
withdrawing group” is not provided. Rather, the footnote associated with both of these deactivating
re
381 features states… “Excludes carboxylic acid and aryl (counted separately), and ketone (conflicting data).
lP

382 Additional electron withdrawing group examples are limited to those described in Cross KP and Ponting

383 DJ, 2021, Developing Structure-Activity Relationships for N-Nitrosamine Activity, Comput Toxicol,
na

384 20:100186, where they are referred to as “β-carbon electron withdrawing groups.” However, as
ur

385 discussed in Cross and Ponting19, the strength of the electron-withdrawing groups also matters, with
Jo

386 examples provided of two strong electron-withdrawing groups negating carcinogenic potential

387 altogether such that even being placed into the highest CPCA category with a limit of 1500 ng/day

388 would be an highly conservative AI. Furthermore, there is some overlap with the activating, benzylic,

389 feature – both act via electron conjugation, and it is important to note that some heterocyclic aromatic

390 rings can be extremely electron-withdrawing in this context yet are currently matching the activating

391 feature.

392 An expansion of understanding for the features that are electronically-conjugated to the NA,

393 whether withdrawing, donating, or electronically neutral but able to delocalise, and thus stabilise,

394 charged intermediates (e.g. the allyl group16,18) would also be valuable. This can include the additional

395 electron-withdrawing groups mentioned by Remya and Suresh33 but not mentioned by Cross and

17
 Business Use

396 Ponting19. The lowest-hanging fruit, however, would be to include as electron-withdrawing groups

397 anything multiply-bonded to a heteroatom (except aromatic systems which, pending the discussion

398 above, match the activating benzylic feature). This includes such groups as phosphonates (P=O),

399 probably both as PIII and PV, but could also include some of the more unusual electronegative

400 heteroatoms. Generally speaking, the substitution pattern from an electronic and steric perspective,

401 can strongly influence the activation or deactivation of the NA, so that further investigations such as

402 quantum mechanical and/or molecular modelling, can supplement the default CPCA potency

403 categories.

f
oo
404 β-Methyl groups
405 Another feature that has led to particular discussion, following its inclusion in the CPCA, is the β-

r
406 -p
methyl group. This is a feature that is reportedly expected to increase carcinogenic potency by
re
407 stabilising the diazonium ion that is formed via induction from the branched carbons1. It is worth
lP

408 noting here the possibility that it may also be due to the Wagner-Meerwein rearrangement34, a facile

409 1,2-sigmatropic shift which can yield a more-substituted, and thus more-stabilised carbocation.
na

410 Compounds in the training set that have led to this observation include 2,6-dimethyl-N-
ur

411 nitrosomorpholine and 3,4,5-trimethyl-1-nitrosopiperazine. It was initially defined as simply “Methyl

Jo

412 group bonded to β-carbon (cyclic or acyclic)” in guidance; however, this definition includes n-propyl

413 and neopentyl groups, as well as matching the 2-hydroxypropyl group which also matches a

414 deactivating feature1. There is however recognition that the definition of this feature should stress

415 the branched nature of the chain since it is the branching that permits the induction, excluding n-

416 propyl, and that neopentyl groups may provide too much steric hindrance1. It was also clarified that,

417 while a heteroatom in the -position is acceptable, as in the case of 2,6-dimethyl-N-

418 nitrosomorpholine, that should be as part of an ongoing chain, excluding 2-halopropyl and 2-

419 hydroxypropyl groups from the feature1. It should be noted that, assuming there is not too much steric

420 hindrance, longer carbon chains may provide a similar effect to that of methyl groups – whichever

18
 Business Use

421 mechanism is applicable - and thus the investigation of additional β-branching, such as ethyl groups,

422 as potentially activating should be investigated.

423 Bioisosteric features

424 Furthermore, there are additional features that are known to impact the potency of NAs that

425 are not included in the current version of the CPCA. This includes, for example, molecular weight,

426 steric hindrance, competing metabolic pathways, and structural elements that are not yet considered,

427 perhaps because they were not in the NAs that were evaluated when the list of activating and

428 deactivating features were described. Nevertheless, this has been acknowledged by the FDA22:

f
oo
429 “The structural environment surrounding the N-nitroso group of the NDSRI is an important factor when

r
430 selecting appropriate reference compounds for a read-across analysis. It may include consideration of

431
-p
the degree of substitution, steric bulk, electronic influences, potential for metabolic activation,
re
432 stability/reactivity of the resulting metabolites, and overall molecular weight.” As an example, the
lP

433 CPCA recognizes the presence of a carboxylic acid anywhere on the molecule as being a strongly

deactivating feature (+3 on the potency score) for pharmacokinetic reasons 16,18,19 but does not yet
na

434

435 recognize other features such as a sulfonic acid that would be expected to have the same impact.
ur

436 These features could however be included in the CPCA in many cases, where a comparable effect on
Jo

437 pharmacokinetics to the acid is expected, i.e. bioisosteres for carboxylic acids, and also similar

438 features, such as the highly-polar sugar-derived chain in N-nitroso meglumine. N-Nitroso meglumine

439 is in class 2 but extremely polar and it can be argued that it should be able to be treated like an acid

440 (+3 points due to polarity rather than merely +1 for the beta-hydroxy). This has been clearly

441 acknowledged by the FDA22: “FDA recognizes that the AI limits recommended by the Agency and those

442 generated by manufacturers and applicants may evolve with advances in the science and generation

443 of data for nitrosamines.” The use of bioisosteres is well established in medicinal chemistry for

444 introducing desirable physicochemical properties35 or optimising binding potential while avoiding

445 particular metabolic or toxic liabilities of the functional group that would typically be used there. The

446 most well-known case of these is the use of tetrazole rings as a bioisiosteric replacement for carboxylic

19
 Business Use

447 acids. A variety of additional bioisosteres for carboxylic acid groups exist36,37, and these should be

448 evaluated for inclusion in the CPCA, with comparable rationale.

449 Pharmacophoric properties

450 Thus far the discussion of features for the CPCA has focussed on specific substructural

451 moieties; however, consideration should also be made of both more complex features, such as

452 pharmacophoric or 3D features and of molecular properties such as the molecular weight25 and logP,

453 long known to affect the biological fate – and thus in vitro and in vivo toxicological response – of

454 chemicals. Recent work25 reminds that the mechanism by which these compounds are mutagenic is

f
oo
455 stoichiometric, and the fewer the NA molecules in a given environment the lower the expected

456 maximum potency per gram becomes.

r
457
-p
This does, however, require some clarity as to the applicability of the CPCA. The workflow can
re
458 assign a category to all nitrosated secondary amines, and is explicit in that, excluding only nitrosated
lP

459 ureas and similar structures where the α-carbon is directly double bonded to a heteroatom and
na

460 nitrosated aromatic nitrogens. However, this category therefore includes both the small molecules

461 with toxicity data (e.g. NDEA, nitrosomorpholine, nitrosoethylisopropylamine, NMBA and NDIPA – to
ur

462 select one from each category) and the larger NDSRIs. While the small-molecule data is necessary for
Jo

463 assessment of potency and thus development of the CPCA limits, indications from within the small-

464 molecule dataset indicate that as size increases potency is reduced38 (by more than the stoichiometric

465 nature25 of the molecular initiating event would imply).

466

467 NAs in category 5

468 Although many of the potential N-nitroso compounds that cannot form diazonium ions (such

469 as nitrosated hydroxylamines, as described by Ponting and Foster17) are outside the scope of the CPCA,

470 there are compounds within the scope of the CPCA that cannot, or do not, α-hydroxylate and thus

471 cannot, or do not, form a diazonium ion. This includes those identified by the first of the questions

472 asked to put a compound into category 5 – whether it has any α-hydrogens at all. α-Hydroxylation is

20
 Business Use

473 thus impossible, and a diazonium ion cannot be formed. This could be readily demonstrated by simple

474 in silico and in vitro metabolism assessments, which would support the position that NAs matching

475 this rule should be outside the scope of the cohort of concern 17, which has implications for the

476 applicability domain of the CPCA. While the CPCA permits the same limit for these category 5

477 compounds as the general TTC, excluding them from the cohort of concern and CPCA entirely would

478 allow the use of less-than-lifetime approaches (fully, rather than partially and temporarily as currently

479 by EMA) and simplify the case of multiple NAs in the same drug product. It is noted that additional

480 research is ongoing to evaluate the applicability of the ICH M7 approach for less-than-lifetime

f
oo
481 exposures more broadly to NAs (i.e., for all potency bins). In the meantime, a non-mutagenic (in the

482 enhanced Ames test) category 5 compound (considering now all three of the structural patterns that

r
483
-p
match this category; although for the second two patterns there are α-hydrogens that can be
re
484 activated, such as the weak carcinogen NDIPA, the potency is expected to be exceptionally low) could
lP

485 also be able to be treated as a non-mutagenic impurity under ICH M7, i.e. ICH Q3A/B limits39,40 rather

486 than 1500 ng/day. Ultimately, the difference between categories 5 and 4 should be that category 5 is
na

487 manageable under standard M7 methods – whether toxicity studies are positive or negative - and
ur

488 should not be conflated with the rest of the NA impurities. These compounds are not likely to be
Jo

489 potent rodent carcinogens – and most indeed are negative either in in vivo genotoxicity studies or

490 indeed rodent carcinogenicity studies, and when not so universally have TD50 values that are above

491 the TTC – whereas category 4 are the lowest potency compounds within the boundaries of the cohort

492 and thus the additional regulatory requirements applicable to NAs. This could be achieved by

493 modifying the cohort definition as has recently been proposed17, and modifying the CPCA flowchart

494 to indicate that category 5 is of even lower concern than category 4.

495

496 Applicability of the CPCA to small NAs

497 Within category 5, there are also compounds which do have α-hydrogens, such as nitroso-

498 diisopropylamine (NDIPA). This compound was initially assigned a limit by read-across to NDEA – the

21
 Business Use

499 prototypical category 1 NA – but data from Druckrey et al41 can be used to assign a TD50 value of 3.84

500 mg/kg/day. This, for what is almost certain to be the most potent of all category 5 NAs (being the

501 smallest and as a result least-hindered), stresses that the assessment of this category at 1500 ng/day

502 is conservative. As has now been performed during the preparation of this manuscript by Health

503 Canada21 (31st May 2024), EMA (4th July)15 and US F.D.A. (4th September 2024)22, a general re-

504 assessment of previously published regulatory limits for compounds such as this, and the category 4

505 nitrosomethylbutylamine (NMBA), in the context of the CPCA, would provide a better estimate of risk.

506

f
oo
507 Beyond the CPCA - chemicals and classes for which data can support a more robust AI
508 than the default CPCA limit

r
509

510
-p
Table 2 shows some compounds and chemical classes, discussed in the supporting information
re
511 (section S2) in more detail, for which the CPCA limits could perhaps be challenged on a class-based
lP

512 basis rather than requiring case-by-case analysis.

na

513 Table 2: Categories of compounds for which more precise limits than the CPCA can be proposed. a can be further
514 modified by the presence of deactivating features; if activating features are present these should be evaluated
515 individually. b Also a nitrosopiperidine. c Alkyl groups of 2-4 carbon chain length; excludes N-nitrosobetahistine
ur

516 and NMPEA.

Jo

Compound/class CPCA class Issues Proposed Rationale

limit
‘pril ACE inhibitors 5 Negative in ICH Q3a/b Extension to
vivo whole class of
mutagenicity arguments
data exists for accepted for
some in class42 quinapril and
ramipril42
Nitrosohydrochlorothiazide 3 Negative in ICH Q3a/b Unstable
vivo, nitrosamine
compound leading to no
unstable in exposure15,21
water
Nitrosoproline 4 Robust ICH Q3a/b Quantitatively
negative excreted, no
carcinogenicity evidence of
data exists carcinogenicity
Nitrosopiperidines, e.g. 3- 3 Robust read- 1300 ng/daya Read across to N-
aminopiperidine across nitrosopiperidine
analogue exists

22
 Business Use

Nitrosomethylphenidateb 4 Some negative 6700 A lack of

carcinogenicity ng/day43 response at this
data exists, dose indicates
though not that any TD50 if
fully robust positive at higher
doses would be
higher than this38
N-Methyl-N-alkylaryl 1 Robust read- 100 or 170 Read-across to
nitrosaminesc across ng/daya NNK (harmonic
analogue exists mean or most
sensitive site)
Diaryl nitrosamines 5 Diazonium 78000 ng/day Not in cohort due
formation to lack of α-
mechanism hydrogen. Read-
impossible across to NDPhA

f
517 Negative in vivo data

oo
518 Some compounds, that would be in category 5, are designated as non-mutagenic impurities (NMI)

r
519 by some regulators15,21 because they have tested negative in in vivo mutagenicity studies and can be

520
-p
controlled according to ICH Q3B. These include N-nitroso-azaerythromycin, N-nitroso-desmethyl-
re
521 azithromycin, N-nitroso-hydrochlorothiazide (HCTZ), and N-nitroso-quinapril. Other regulators22 have
lP

522 set acceptable intakes for these compounds, based on CPCA only (category 4/5 compounds with limits
na

523 of 1500 ng/day), and no guidance has been provided how the negative in vivo mutagenicity data that

524 has been generated for these substances should be used. This essentially indicates that these
ur

525 regulators are defaulting to the use of CPCA rather than leveraging robust experimental data. An
Jo

526 alternate possibility is that these limits are achievable by manufacturers based on process capability

527 and thus it is considered they may be sufficient, but this leads to “spec-setting” analogously to the

528 above, where 1500 ng/day is achieved by quality and therefore there is no incentive to allow the

529 higher levels – this is not just a theoretical concern; varying processes between companies and

530 between products within companies (for HCTZ, in particular, the unstable nature of the NA may mean

531 that the difference is between analytical methods, not true levels), can mean that later-reported

532 products cannot achieve the CPCA limit, but are no less safe above it. A quality-derived specification

533 should not supersede data-based risk assessment for limit derivation.

23
 Business Use

534 In cases such as these, where there is some evidence that the compounds are not of significant

535 mutagenic potential, or indeed any mutagenic effect in vivo at all, it is not clear what value is added

536 by publishing CPCA-based limits. This is especially significant given that the evidence has been

537 accepted by other health authorities, leading to inter-agency variation that creates significant

538 problems for manufacturers. Leaving a limit as pending, as was done for nitroso-ciprofloxacin for

539 several months by the FDA22, would be preferable – giving notice that the agency is concerned about

540 the impurity, but giving time for compound-specific risk assessment and/or global harmonisation via

541 the inter-agency Nitrosamines International Technical Working Group (NITWG)1 to occur. This would

f
oo
542 prevent a need for the limit to change back and forth as data is generated on both the toxicity and

543 observed levels of the NA in drug product. It is particularly concerning that these published values

r
544
-p
can be subject to misinterpretation and taken to represent an actual safety limit rather than how they
re
545 should more appropriately be viewed, which is as a default tier-0 limit that is a useful default in the
lP

546 absence of data to support a chemical-specific AI.

na

547 Why conservative regulatory default limits can be problematic if treated as definitive
548 safety limits
ur

549 The introduction of CPCA marks a significant change in HAs approach to defining AI limits for NA
Jo

550 impurities. Previously, in the absence of any guiding principles, the process relied on HAs to derive the

551 AI limits on a case-by-case basis. This led to derivation of AIs for specific NDSRIs that were unclear and

552 often divergent to limits proposed by the applicant. In addition to delays in receiving communication

553 of this (due to the associated additional workload), the result of this case-by-case approach could be

554 significant, especially where AIs proposed by regulators are markedly lower than those proposed by

555 the applicant (e.g. the case of nitrosovarenicline described by Ponting et al16). With the advent of the

556 CPCA, limits can now be determined directly by the pharmaceutical manufacturers with confidence

557 and without the need to obtain regulatory endorsement provided the CPCA rules are correctly applied.

558 However, it was recently demonstrated that approximately 65 % of NDSRIs derived from secondary

559 amines will fall under the least potent categories 4 and 5 (AI 1500 ng/day) and another 15 % under

24
 Business Use

560 the medium potency category 3 (AI 400 ng/day).7 As noted already the CPCA remains highly

561 conservative indeed the data described implies that ca. 20 % remain in categories 1 (AI 18 ng/day) and

562 2 (AI 100 ng/day), i.e., their potency is assumed to be comparable with NDEA and NDMA, respectively.

563 This poses a significant challenge for the pharmaceutical industry, because the application of

564 reasonable and scientifically sound principles such as the adjustment for less-than-lifetime exposure44

565 or scaling to molecular weight,25 which would allow to increase those limits where justified, is

566 currently not accepted. Figure 4 seeks to illustrate how molecular weight scaling and less-than-lifetime

567 adjustment would lift the LOQs required for proving absence of 27 NDSRIs belonging to CPCA

f
oo
568 categories 1 and 2.

r
569 For an NDSRI, unless it is formed from an API impurity and not the API itself, the formation clearly

570
-p
cannot be mitigated by reducing the content of the vulnerable amine. Two options remain, namely,
re
571 reformulation with a nitrosation scavenger45 and/or reduction of the nitrite content by selecting
lP

572 alternative excipient suppliers, which may or may not be available.8 Option 1 requires formulation
na

573 development, stability testing and a bioequivalence study. Adding to this the subsequent regulatory

574 submission as a major variation and approval, collectively has a timeline of several years, particularly
ur

575 if different markets where different Health Authorities are concerned. Option 2 relies on the
Jo

576 availability of low-nitrite excipient(s), and even then, a base level of nitrosation may remain, caused

577 by NOx species from the air. The latter was recently demonstrated by Fukuda et al. for a metformin

578 drug product.9 As per the current guidance, this will require manufacturers to establish a full testing

579 regime at release, unless they can demonstrate that the NDSRI is consistently below 10 % (to avoid

580 skip testing) or below 30 % (to avoid full testing) of the AI limit, 6 under the assumption that a patient

581 will always consume the full maximum daily dose.

582 It was already demonstrated before that this burden of evidence will be impossible to achieve in many

583 cases, as it requires analytical quantification limits beyond current technical feasibility, especially for

584 a routine GMP laboratory. Many companies will rely heavily on contract labs providing the required

25
 Business Use

585 expert skills and instrumentation for these activities, and it is uncertain whether the available

586 capacities will be sufficient to cover the demand. Small companies and manufacturers of generics may

587 not be able or willing to invest in mitigation actions or routine testing at all and instead prune affected

588 products, decreasing competition in the market and increasing the prices for the remaining products.

589 At the opposite end of the product lifecycle, highly conservative AI limits can become an obstacle for

590 the development of new drugs, as the avoidance of secondary amines in the API and relevant starting

591 materials and intermediates reduces the available design space significantly.2 The use of highly

592 conservative AIs can be particularly challenging given that the ICH M7 less-than-lifetime approach,

f
oo
593 which is intended to support higher limits during the drug development process, is currently not

594 accepted for NA impurities.

r
-p
re
lP
na
ur
Jo

595
596 Figure 42: Analytical challenges for CPCA category 1&2 compounds and how they could be partially mitigated by
597 allowing molecular weight and LTL correction. A) Maximum daily doses B) Required LOQ to prove absence of the
598 NA, no MW scaling or LTL correction C) Number of API molecules per NDSRI molecule at the required LOQ level,

26
 Business Use

599 no MW scaling or LTL correction D) Required LOQ to prove absence of the NA if MW scaling and LTL correction
600 were allowed. The dashed lines represent 100 ppb, 10 ppb and 1 ppb in terms of required analytical sensitivity,
601 which is generally considered possible, difficult or almost impossible to achieve, respectively.

602

603 Conclusions
604
605 The introduction of the CPCA was a pivotal milestone in the assessment and control of NA

606 impurities providing a framework for the management of NDSRIs in drug products. It achieves its

607 design intent: to define conservative but harmonised default AI limits for NDSRI’s in drug products

608 that can be used by HAs and MAHs, in the absence of any associated compound-specific data. It builds

f
609 on regulatory experience from a number of different fields, providing a tiered TTC-like approach which

oo
610 allows the rapid screening of all NDSRIs and subsequent assignment of an associated AI limit. This

r
611 -p
provides a conservative starting point for the assessment of additional toxicology data - in vitro or in
re
612 vivo as appropriate – or selection of suitable read-across analogues, in order to provide a robust

613 compound specific limit, rather than being a final safety threshold at levels above which market action
lP

614 is necessary. It is important to recognise that default CPCA derived AI limits are the result of an initial
na

615 default Tier-0 screening approach that are associated with a de minimis risk and so would not
ur

616 necessarily a reflect a definitive position on the safety of an NA impurity.

Jo

617 The consequences of highly conservative limits are significant and can range from increased

618 resources to account for additional quality measures, and potentially to batch withdrawals or

619 complete withdrawals of a drug from a market if the limit is unachievable.

620 It is important to note that the CPCA is expected to evolve as further QSAR research is

621 undertaken - by health authorities, academia, industry, and QSAR vendors – and all of the above acting

622 in concert under the auspices of the Health and Environmental Sciences Institute (HESI). The evolution

623 of the CPCA will refine the classes, increasing the accuracy, and is expected to change the default AIs

624 for some NDSRIs. This is consistent with the iterative nature of human health risk assessment where

625 screening values, while designed to be very conservative, can play a valuable role in ensuring an

626 efficient use of resources. Further evaluation of existing data (e.g., to support read across) and/or the

27
 Business Use

627 generation of additional safety data can often support higher exposure limits without compromising

628 patient safety.

629
630 Acknowledgements
631 The authors would like to acknowledge Dr Adrian Fowkes (Lhasa Limited) and Dr Michael W

632 Urquhart (GSK) for valuable comments during the preparation of this manuscript, as well as Dr Gair

633 Ford (AZ) for curation of the data captured in Figure 1

634 Conflict of interest

635 Two authors are employed by Lhasa Limited, who market Derek Nexus referred to. The

f
636 authors’ personal experiences in using published CPCA categorisation in the development of Derek

oo
637 Nexus are cited as an example of the utility of these, but this reference to Derek Nexus is intended to
638 be illustrative rather than exhaustive and it is acknowledged that comparable tools from other vendors

r
639 exist. All other authors report relationships with their listed affiliations respectively that includes:
640 employment and equity or stocks. -p
re
641

642
lP

643 References
644
na

645 (1) Kruhlak, N. L.; Schmidt, M.; Froetschl, R.; Graber, S.; Haas, B.; Horne, I.; Horne, S.; King, S. T.;
646 Koval, I. A.; Kumaran, G.; Langenkamp, A.; McGovern, T. J.; Peryea, T.; Sanh, A.; Ferreira, A. S.;
ur

647 Van Aerts, L.; Vespa, A.; Whomsley, R. Determining Recommended Acceptable Intake Limits for
648 N-Nitrosamine Impurities in Pharmaceuticals: Development and Application of the
Jo

649 Carcinogenic Potency Categorization Approach (CPCA). Regul. Toxicol. Pharmacol. 2024,
650 105640. https://doi.org/10.1016/j.yrtph.2024.105640.
651 (2) Nudelman, R.; Kocks, G.; Mouton, B.; Ponting, D. J.; Schlingemann, J.; Simon, S.; Smith, G. F.;
652 Teasdale, A.; Werner, A.-L. The Nitrosamine “Saga”: Lessons Learned from Five Years of
653 Scrutiny. Org. Process Res. Dev. 2023, 27 (10), 1719–1735.
654 https://doi.org/10.1021/acs.oprd.3c00100.
655 (3) Schlingemann, J.; Burns, M. J.; Ponting, D. J.; Avila, C. M.; Romero, N. E.; Jaywant, M. A.; Smith,
656 G. F.; Ashworth, I. W.; Simon, S.; Saal, C.; Wilk, A. The Landscape of Potential Small and Drug
657 Substance Related Nitrosamines in Pharmaceuticals. J. Pharm. Sci. 2023, 112, 1287–1304.
658 https://doi.org/10.1016/J.XPHS.2022.11.013.
659 (4) Li, X.; Le, Y.; Seo, J. E.; Li, Y.; Guo, X.; Chen, S.; Mittelstaedt, R.; Moore, N.; Guerrero, S.; Sims,
660 A.; Atrakchi, A.; McGovern, T. J.; Davis-Bruno, K. L.; Keire, D. A.; Elespuru, R. K.; Heflich, R. H.;
661 Mei, N. Revisiting the Mutagenicity and Genotoxicity of N-Nitroso Propranolol in Bacterial and
662 Human in Vitro Assays. Regul. Toxicol. Pharmacol. 2023, 141, 105410.
663 https://doi.org/10.1016/j.yrtph.2023.105410.
664 (5) Thomas, R.; Thresher, A.; Ponting, D. J. Utilisation of Parametric Methods to Improve
665 Percentile-Based Estimates for the Carcinogenic Potency of Nitrosamines. Regul. Toxicol.
666 Pharmacol. 2021, 121, 104875. https://doi.org/10.1016/j.yrtph.2021.104875.

28
 Business Use

667 (6) European Medicines Agency (EMA). Procedure under Article 5(3) of Regulation EC (No)
668 726/2004: Nitrosamine Impurities in Human Medicinal Products. Procedure Number:
669 EMEA/H/A-5(3)/1490. EMA/369136/2020; 2020.
670 (7) Burns, M. J.; Ponting, D. J.; Foster, R. S.; Thornton, B. P.; Romero, N. E.; Smith, G. F.; Ashworth,
671 I. W.; Teasdale, A.; Simon, S.; Schlingemann, J. Revisiting the Landscape of Potential Small and
672 Drug Substance Related Nitrosamines in Pharmaceuticals. J. Pharm. Sci. 2023, 112 (12), 3005–
673 3011. https://doi.org/10.1016/j.xphs.2023.10.001.
674 (8) Boetzel, R.; Schlingemann, J.; Hickert, S.; Korn, C.; Kocks, G.; Luck, B.; Blom, G.; Harrison, M.;
675 Franc, M.; Allain, L.; Wu, Y. A Nitrite Excipient Database : A Useful Tool to Support N-
676 Nitrosamine Risk Assessments for Drug Products. J. Pharm. Sci. 2022.
677 https://doi.org/10.1016/j.xphs.2022.04.016.
678 (9) Fukuda, S.; Kondo, K.; Fukumoto, S.; Takenaka, N.; Uchikawa, O.; Yoshida, I. N-
679 Nitrosodimethylamine Formation in Metformin Drug Products by the Reaction of
680 Dimethylamine and Atmospheric NO2. Org. Process Res. Dev. 2023, 27 (11), 2123–2133.
681 https://doi.org/10.1021/acs.oprd.3c00274.

f
682 (10) Peto, R.; Gray, R.; Brantom, P.; Grasso, P. Effects on 4080 Rats of Chronic Ingestion of N-

oo
683 Nitrosodiethylamine or N-Nitrosodimethylamine: A Detailed Dose-Response Study. Cancer Res.
684 1991, 51, 6415–6451.
685 (11) Kroes, R.; Renwick, A. G.; Cheeseman, M. A.; Kleiner, J.; Mangelsdorf, I.; Piersma, A.; Schilter,

r
686 B.; Schlatter, J.; Van Schothorst, F.; Vos, J. G.; Würtzen, G. Structure-Based Thresholds of
687
688
-p
Toxicological Concern (TTC): Guidance for Application to Substances Present at Low Levels in
the Diet. Food Chem. Toxicol. 2004, 42 (1), 65–83. https://doi.org/10.1016/j.fct.2003.08.006.
re
689 (12) Kroes, R.; Galli, C.; Munro, I.; Schilter, B.; Tran, L. A.; Walker, R.; Würtzen, G. Threshold of
690 Toxicological Concern for Chemical Substances Present in the Diet: A Practical Tool for
lP

691 Assessing the Need for Toxicity Testing. Food Chem. Toxicol. 2000, 38 (2–3), 255–312.
692 https://doi.org/10.1016/S0278-6915(99)00120-9.
693 (13) Kroes, R.; Kleiner, J.; Renwick, A. The Threshold of Toxicological Concern Concept in Risk
na

694 Assessment. Toxicol. Sci. 2005, 86 (2), 226–230. https://doi.org/10.1093/toxsci/kfi169.

695 (14) ICH. M7(R2): Assessment and Control of DNA Reactive (Mutagenic) Impurities in
696 Pharmaceuticals to Limit Potential Carcinogenic Risk; 2023.
ur

697 (15) European Medicines Agency (EMA). Questions and Answers for Marketing Authorisation
698 Holders/Applicants on the CHMP Opinion for the Article 5(3) of Regulation (EC) No 726/2004
Jo

699 Referral on Nitrosamine Impurities in Human Medicinal Products. EMA/409815/2020 Rev. 21

700 and Appendices.; 2024.
701 (16) Ponting, D. J.; Dobo, K. L.; Kenyon, M. O.; Kalgutkar, A. S. Strategies for Assessing Acceptable
702 Intakes for Novel N-Nitrosamines Derived from Active Pharmaceutical Ingredients. J. Med.
703 Chem. 2022, 65 (23), 15584–15607. https://doi.org/10.1021/ACS.JMEDCHEM.2C01498.
704 (17) Ponting, D. J.; Foster, R. S. Drawing a Line: Where Might the Cohort of Concern End? Org.
705 Process Res. Dev. 2023, 27 (10), 1703–1713. https://doi.org/10.1021/acs.oprd.3c00008.
706 (18) Thomas, R.; Tennant, R. E.; Oliveira, A. A. F.; Ponting, D. J. What Makes a Potent Nitrosamine?
707 Statistical Validation of Expert-Derived Structure Activity Relationships. Chem. Res. Toxicol.
708 2022, 11, 1997–2013. https://doi.org/10.1021/acs.chemrestox.2c00199.
709 (19) Cross, K. P.; Ponting, D. J. Developing Structure-Activity Relationships for N-Nitrosamine
710 Activity. Comput. Toxicol. 2021, 20, 100186. https://doi.org/10.1016/J.COMTOX.2021.100186.
711 (20) Dobo, K. L.; Kenyon, M. O.; Dirat, O.; Engel, M.; Fleetwood, A.; Martin, M.; Mattano, S.; Musso,
712 A.; McWilliams, J. C.; Papanikolaou, A.; Parris, P.; Whritenour, J.; Yu, S.; Kalgutkar, A. S.
713 Practical and Science-Based Strategy for Establishing Acceptable Intakes for Drug Product N ‑
714 Nitrosamine Impurities. Chem. Res. Toxicol. 2022, 35, 475–489.
715 https://doi.org/10.1021/ACS.CHEMRESTOX.1C00369.
716 (21) Health Canada. Guidance on Nitrosamine Impurities in Medications. Revision as of 31st May
717 2024; 2024.

29
 Business Use

718 (22) United States Food and Drug Administration (U.S. FDA). Recommended Acceptable Intake
719 Limits for Nitrosamine Drug Substance-Related Impurities (NDSRIs). Revision as of 4th
720 September 2024.
721 (23) Snodin, D. J. Mutagenic Impurities in Pharmaceuticals: A Critical Assessment of the Cohort of
722 Concern with a Focus on N-Nitrosamines. Regul. Toxicol. Pharmacol. 2023, 141, 105403.
723 https://doi.org/10.1016/j.yrtph.2023.105403.
724 (24) Thresher, A.; Foster, R. S.; Ponting, D. J.; Stalford, S. A.; Tennant, R. E.; Thomas, R. Are All
725 Nitrosamines Concerning? A Review of Mutagenicity and Carcinogenicity Data. Regul. Toxicol.
726 Pharmacol. 2020, 116 (June), 104749. https://doi.org/10.1016/j.yrtph.2020.104749.
727 (25) Fine, J.; Allain, L.; Schlingemann, J.; Ponting, D. J.; Thomas, R.; Johnson, G. E. Nitrosamine
728 Acceptable Intakes Should Consider Variation in Molecular Weight: The Implication of
729 Stoichiometric DNA Damage. Regul. Toxicol. Pharmacol. 2023, 145, 105505.
730 https://doi.org/10.1016/j.yrtph.2023.105505.
731 (26) Ponting, D. J.; Burns, M. J.; Foster, R. S.; Hemingway, R.; Kocks, G.; MacMillan, D. S.; Shannon-
732 Little, A. L.; Tennant, R. E.; Tidmarsh, J. R.; Yeo, D. J. Use of Lhasa Limited Products for the In

f
733 Silico Prediction of Drug Toxicity. In Methods in Molecular Biology; Benfenati, E., Ed.; Humana

oo
734 Press Inc., 2022; Vol. 2425, pp 435–478. https://doi.org/10.1007/978-1-0716-1960-
735 5_17/COVER.
736 (27) Shapiro, S. S.; Wilk, M. B. An Analysis of Variance Test for Normality (Complete Samples).

r
737 Biometrika 1965, 52 (3–4), 591–611. https://doi.org/10.1093/biomet/52.3-4.591.
738
739
-p
(28) Science and Judgment in Risk Assessment; Committee on risk assessment of hazardous air
pollutants, Ed.; National academy press: Washington, 1994.
re
740 (29) United States Food and Drug Administration (U.S. FDA). Control of Nitrosamine Impurities in
741 Human Drugs, Guidance for Industry; 2021.
lP

742 (30) Munro, I. C.; Ford, R. A.; Kennepohl, E.; Sprenger, J. G. Correlation of Structural Class with No-
743 Observed-Effect Levels: A Proposal for Establishing a Threshold of Concern. Food Chem.
744 Toxicol. 1996, 34 (9), 829–867. https://doi.org/10.1016/S0278-6915(96)00049-X.
na

745 (31) European Food Safety Authority and World Health Organization. Review of the Threshold of
746 Toxicological Concern (TTC) Approach and Development of New TTC Decision Tree. EFSA
747 Support. Publ. 2016, 13 (3). https://doi.org/10.2903/sp.efsa.2016.EN-1006.
ur

748 (32) Health and Environmental Sciences Institute (HESI). HESI GTTC Launches New (Q)SAR/QM
749 Nitrosamines Working Group. https://connect.hesiglobal.org/Amd56N (accessed 2024-04-28).
Jo

750 (33) Remya, G. S.; Suresh, C. H. Quantification and Classification of Substituent Effects in Organic
751 Chemistry: A Theoretical Molecular Electrostatic Potential Study. Phys. Chem. Chem. Phys.
752 2016, 18 (30), 20615–20626. https://doi.org/10.1039/c6cp02936a.
753 (34) Hanson, J. R. Wagner–Meerwein Rearrangements. In Comprehensive Organic Synthesis;
754 Elsevier, 1991; pp 705–719. https://doi.org/10.1016/B978-0-08-052349-1.00077-9.
755 (35) Meanwell, N. A. Synopsis of Some Recent Tactical Application of Bioisosteres in Drug Design. J.
756 Med. Chem. 2011, 54 (8), 2529–2591. https://doi.org/10.1021/jm1013693.
757 (36) Lassalas, P.; Gay, B.; Lasfargeas, C.; James, M. J.; Tran, V.; Vijayendran, K. G.; Brunden, K. R.;
758 Kozlowski, M. C.; Thomas, C. J.; Smith, A. B.; Huryn, D. M.; Ballatore, C. Structure Property
759 Relationships of Carboxylic Acid Isosteres. J. Med. Chem. 2016, 59 (7), 3183–3203.
760 https://doi.org/10.1021/acs.jmedchem.5b01963.
761 (37) Winters, M. P.; Crysler, C.; Subasinghe, N.; Ryan, D.; Leong, L.; Zhao, S.; Donatelli, R.; Yurkow,
762 E.; Mazzulla, M.; Boczon, L.; Manthey, C. L.; Molloy, C.; Raymond, H.; Murray, L.; McAlonan, L.;
763 Tomczuk, B. Carboxylic Acid Bioisosteres Acylsulfonamides, Acylsulfamides, and Sulfonylureas
764 as Novel Antagonists of the CXCR2 Receptor. Bioorg. Med. Chem. Lett. 2008, 18 (6), 1926–
765 1930. https://doi.org/10.1016/j.bmcl.2008.01.127.
766 (38) Felter, S. P.; Ponting, D. J.; Mudd, A. M.; Thomas, R.; Oliveira, A. A. F. Maximizing Use of
767 Existing Carcinogenicity Data to Support Acceptable Intake Levels for Mutagenic Impurities in

30
 Business Use

768 Pharmaceuticals: Learnings from N-Nitrosamine Case Studies. Regul. Toxicol. Pharmacol. 2023,
769 143, 105459. https://doi.org/10.1016/j.yrtph.2023.105459.
770 (39) ICH. Q3A(R2): Impurities in New Drug Substances; 2006.
771 (40) ICH. Q3B (R2): Impurities in New Drug Products; 2006.
772 (41) Druckrey, H.; Preussmann, R.; Ivankovic, S.; Schmähl, D.; Afkham, J.; Blum, G.; Mennel, H. D.;
773 Müller, M.; Petropoulos, P.; Schneider, H. Organotrope Carcinogene Wirkungen Bei 65
774 Verschiedenen N-Nitroso-Verbindungen an BD-Ratten. Z. Für Krebsforsch. 1967, 69 (2), 103–
775 201. https://doi.org/10.1007/BF00524152.
776 (42) Cheung, J.; Dobo, K.; Zhang, S.; Nudelman, R.; Schmidt, F.; Wenzel, J.; Czich, A.; Schuler, M.
777 Evaluation of the Nitrosamine Impurities of ACE Inhibitors Using Computational, in Vitro, and in
778 Vivo Methods Demonstrate No Genotoxic Potential. Environ. Mol. Mutagen. 2024, 67, 203–
779 221. https://doi.org/10.1002/em.22618.
780 (43) Giner-Sorolla, A.; Greenbaum, J.; Last-Barney, K.; Anderson, L. M.; Budinger, J. M. Lack of
781 Carcinogenic Effect of Nitrosochlordiazepoxide and of Nitrosomethylphenidate given Orally to
782 Mice. Food Cosmet. Toxicol. 1980, 18 (1), 81–83. https://doi.org/10.1016/0015-

f
783 6264(80)90015-2.

oo
784 (44) Bercu, J. P.; Masuda-Herrera, M.; Johnson, G. E.; Czich, A.; Glowienke, S.; Kenyon, M.; Thomas,
785 R.; Ponting, D. J.; White, A.; Cross, K.; Waechter, F.; Rodrigues, M. A. C. Use of Less-than-
786 Lifetime ( LTL ) Durational Limits for Nitrosamines : Case Study of N -Nitrosodiethylamine (

r
787 NDEA ). Regul. Toxicol. Pharmacol. 2021, 123 (April), 104926.
788
789 (45)
-p
https://doi.org/10.1016/j.yrtph.2021.104926.
Homšak, M.; Trampuž, M.; Naveršnik, K.; Kitanovski, Z.; Žnidarič, M.; Kiefer, M.; Časar, Z.
re
790 Assessment of a Diverse Array of Nitrite Scavengers in Solution and Solid State: A Study of
791 Inhibitory Effect on the Formation of Alkyl-Aryl and Dialkyl N-Nitrosamine Derivatives.
lP

792 Processes 2022, 10 (11), 2428. https://doi.org/10.3390/pr10112428.

793
na
ur
Jo

31
The CPCA provides a tiered TTC-like triage for nitrosamine impurity assessment.

de minimis risk TTC-like limits should not be definitive safety thresholds.

Risk assessment is an iterative process – CPCA refinements are possible.

Information from close analogues is an important part of risk assessment.

Compound-specific data, from in silico to in vivo studies, refines initial triage.

of
ro
-p
re
lP
na
ur
Jo
All authors are employed by their respective affiliations.

f
r oo
-p
re
lP
na
ur
Jo
Declaration of interests

☐ The authors declare that they have no known competing financial interests or personal relationships
that could have appeared to influence the work reported in this paper.

☒ The authors declare the following financial interests/personal relationships which may be considered
as potential competing interests:

David J Ponting, Robert Thomas report financial support was provided by Lhasa Limited. Andreas
Czich reports financial support was provided by Sanofi-Aventis Deutschland GmbH. Susan Felter
reports financial support was provided by The Procter & Gamble Company. Susanne Glowienke
reports financial support was provided by Novartis Pharma AG. James S Harvey reports financial

of
support was provided by GSK. Raphael Nudelman reports financial support was provided by Teva
Pharmaceutical Industries Ltd. Joerg Schlingemann, Stephanie Simon report financial support was

ro
provided by Merck Healthcare KGaA. Graham F Smith, Andrew Teasdale report financial support was
provided by AstraZeneca UK Limited. Two authors [DJP, RT] are employed by Lhasa Limited, who

-p
market Derek Nexus referred to. The authors’ personal experiences in using published CPCA
categorisation in the development of Derek Nexus are cited as an example of the utility of these, but
re
this reference to Derek Nexus is intended to be illustrative rather than exhaustive and it is
acknowledged that comparable tools from other vendors exist. If there are other authors, they
lP

declare that they have no further known competing financial interests or personal relationships that
could have appeared to influence the work reported in this paper.
na
ur
Jo