R.A. Jolly et al.

Regulatory Toxicology and Pharmacology 152 (2024) 105672

Fig. 6b. Mutation data – duodenum.
Table 4
NOEL, BMDL, BMD and BMDLValues for NDEA, NDMA and NDSRIs.

Table 5
CPCA AI, published21 and alternate AI values32 for comparator nitrosamines and
NDSRIs.

Agent

Liver
NOEL
(mg/
kg)

Liver
BMDL
(mg/
kg)

Liver
BMD
(mg/
kg)

Liver
BMDU
(mg/
kg)

TD50*
(mg/
kg)

Reference

NDEA
NDEA

0.09
0.1

0.022
0.1

NR
NR

NR
1

0.0265
0.0265

NDMA

0.36

0.21

0.32

0.46

0.096

NFLX
NDLX
NATX

5
5
5

6.3
6.8
4.4

11.4
13
8.7

24.9
24.7
29.3

ND
ND
ND

Akai (2015)
Bercu et al. (2023a,
b)
Gollupadi (1998);
Johnson et al.
(2021); Lynch et al.
(2024)
Current paper
Current paper
Current paper

Nitrosamine
or NDSRI

AI
based
on
CPCA
(ng/
day)

Published
AIs (ng/
day)

AI based
on Phys
Chem
read
across
(ng/
day)

AI based on
QM
modeling
(ng/day)

AI based on
mutation
frequency
(ng/day)

NDEA

18/
26.5
18/
26.5
18/
26.5
18/
26.5
18/
26.5
18/
26.5

18/26.5

NA

NA

22–100

96

NA

NA

60

100

NA

NA

ND

100

537

1140

>5000

100

537

1152

>5000

100

537

1190

>4400

NDMA
NNK
NFLX

ND = No data available; NR = Not reported.
*TD50 values based on Lhasa Carcinogenicity database.

NDLX
NATX

compared to those potent nitrosamines.
Published regulatory AIs, CPCA AIs, and AIs for these nitrosamines
based on alternate read across and modeling approaches are shown in
Table 5.
The CPCA approach, which is based on the nitrosamine structure
only, would give the lowest-bin AI value for all 3 NDSRIs: 18 or 26.5 ng/
day. The regulatory values published for NFLX, NDLX and NATX are
based on a simple 2D structure read across methodology to NNK. As
shown in the table, newer, non-traditional methods for AI assessment,
such as read across methods using physicochemical properties/QM
modeling/or an AI derived from in vivo mutation data all indicate lower
potencies and substantially higher AI values for the three NDSRIs.

N/A = not applicable. For NDEA, NDMA and NNK the AI are based on carcinogenicity data and these compounds were the basis for the physicochemical
read across approach.

et al., 2020; Dobo et al., 2022; Ponting and Foster, 2023). Thus, AIs for
NDSRIs should be assessed on a case-by-case basis. The read across and
published AIs for NFLX, NDLX and NATX were based on NNK as the
comparator compound which has structural elements in common with
the three NDSRIs evaluated. However, additional analyses for read
across (presented herein) which included the use of physicochemical
properties and/or mechanistic data evaluating bond energy of the
reactive intermediate indicate that the NDSRIs will not be as potent as
NNK. The lower potency of the estimated AIs for these NDSRIs were
corroborated empirically using in vivo mutation frequency data from
transgenic rats administered maximum tolerated doses of each NDSRI.
Current AI-setting strategies in recent guidance from both EMA and
FDA employ read across or structure activity relationships or the CPCA
method to assign NDSRIs to classes (FDA, 2023; EMA, 2023). The
worst-case class used in the CPCA assumes that a NDSRI is as potent as

4. Discussion
Nitrosamines have generally been considered potentially potent
mutagens and carcinogens, warranting special consideration as a Cohort
of Concern (ICH M7 (R2) 2023). However, NDSRIs associated with
pharmaceutical API are generally not expected to be as potent as LMW
nitrosamines such as NDEA or NDMA due to MW, steric hindrance,
mitigating structural features, and competing metabolism (Thresher
8

30 / 34 ページ