Regulatory Toxicology and Pharmacology 152 (2024) 105672

R.A. Jolly et al.

calculated PDE (permitted daily exposure) based on the rodent cancer
bioassay is 1–2 orders of magnitude higher that the PDE based on in vivo
mutagenicity data (Johnson et al., 2021, Table 4). Furthermore, use of a
dose causing no effect (NOEL) as the point of departure is more conservative than using a dose eliciting a 50% tumor rate in rodents (e.g.,
the TD50). In addition, the BMD analysis, also a well-validated approach
to data analysis (Hardy et al., 2017; EPA, 2012), employs all available
data as opposed to a single derived TD50, and a BMDL (95% lower limit)
provides an additional level of conservatism. While the BMDL approach
is more analytically robust than the single-point NOEL value, this
analysis used the lower of the two values (NOEL or BMDL) to derive the
AI as an additional conservative measure.
The NOELs and BMDL values of the NDSRIs relative to the values for
NDMA in Table 4 show that there is at least a 10-100x difference in
potency present and support an argument for the NDSRIs being both less
potent and not in the ICH M7 (R2) Cohort of Concern. A summary of
proposed AI values for the NDSRIs is shown in Table 5 and the in vivo
mutagenicity data suggests that, even using the most conservative
(numerically lowest) estimate of AI, levels as high as 4400 ng/day
derived as described in ICH M7 (R2) would be appropriately protective
based on a threshold argument.
The approach of using the BMDL derived from in vivo mutagenicity
data has been employed to determine a regulatory exposure limit for
ethylmethane sulfonate (Gocke and Müller 2009; Gocke et al., 2009) and
the potent alkylating agents N-methyl-N-nitrosourea (MNU) and
N-ethyl-N-nitrosourea (ENU) (Johnson et al., 2014). Since that time, the
use of BMDL to derive justifiable human exposure limits has been used in
case studies and regulatory submissions with other genotoxic compounds such as benzene, mitomycin C, and bis-chloronitrosourea (e.g.,
Heflich et al., 2020; White et al., 2020; Ponting et al., 2022).
As aforementioned, apart from the clear scientific rationale for
mutagenicity-based assessments to support derivation of AI limits, one
must consider the practicality (and feasibility) of conducting TGR
studies as opposed to rodent carcinogenicity studies for the plethora of
NDSRIs. This approach would decrease use of animals and align with
principles of the 3Rs. The quantitative approach using in vivo genotoxicity data is particularly relevant where carcinogenicity data are unavailable or of poor quality and when in vivo mutagenicity dose-response
data display a mechanistically understood response threshold (COM,
2018; Ponting et al., 2022). The data presented herein show that the
NDSRIs of fluoxetine, duloxetine and atomoxetine are mutagenic in vitro
and in vivo. However, the in vivo mutagenicity data demonstrate that all
the NDSRI compounds exhibit a threshold response with NOEL at 5
mg/kg. To put this dose in perspective, the NOEL of 5 mg NFLX/kg is
more than 3-fold greater than the highest approved maximum dose of
the API, fluoxetine (80 mg or 1.6 mg/kg for a 50-kg adult; USPI).

authorities. Given the potentially severe health risks associated with
nitrosamines, a cautious and comprehensive approach is necessary to
ensure patient safety while maintaining an adequate supply of medicines to treat diseases with significant morbidity.
Disclaimer
Robert A Jolly, Paul D Cornwell, Jessica Noteboom, Fareed B Sayyed,
Bishnu Thapa, and Lorrene A Buckley are employees of and may own
shares in Eli Lilly and Company, Inc.
Funding
All funding for this work was provided by Eli Lilly and Company,
Indianapolis IN, USA.
CRediT authorship contribution statement
Robert A. Jolly: Writing – review & editing, Writing – original draft,
Visualization, Supervision, Methodology, Investigation, Formal analysis, Conceptualization. Paul D. Cornwell: Writing – review & editing,
Supervision, Investigation, Conceptualization. Jessica Noteboom:
Validation, Project administration, Methodology, Data curation. Fareed
Bhasha Sayyed: Writing – original draft, Methodology, Investigation,
Formal analysis, Data curation. Bishnu Thapa: Formal analysis, Data
curation. Lorrene A. Buckley: Writing – review & editing, Writing –
original draft, Project administration.
Declaration of competing interest
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.
Data availability
The data that has been used is confidential.
Acknowledgments
The authors thank David Ackley, Douglas Roepke, Ellen Cannady,
Sheroy Minocherhomji and Prashant Desai for their reviews of the
manuscript and helpful discussions.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.yrtph.2024.105672.

5. Conclusion
The CPCA published by HAs is a good start in the evaluation of unknown NDSRIs and allows for incorporation of additional data to inform
a weight-of-evidence risk assessment. The inclusion of physicochemical
properties and mechanistic modeling are important for an informed
evaluation of risk. In a well-conducted TGR study, a negative result
should serve as qualification of a nitrosamine impurity and positive
results should be used to assess relative potency to LMW comparators.
For TGR-positive mutagenic substances, risk characterization can be
evolved further to derive improved estimates of AI limits based on the in
vivo mutation data.
It should be noted that the current products which can potentially
contain NDSRIs have been on the market for decades, and there has been
no signal for increased cancer risk (e.g. fluoxetine). Despite limited
analytical power, post-marketing pharmacovigilance has not identified
any association between these products and a risk for cancer. This would
indicate that the level of risk of NDSRIs in these products does not justify
the need for the very low AIs currently recommended by health

References
Agilent Technoliges, 2015a. Transpack Packaging Extract For Lambda Transgenic Shuttle
Vector Recovery INSTRUCTION MANUAL Catalog #200220 (400 packaging
reactions), #200221 (100 packaging reactions), and #200223 (50 packaging
reactions) Revision A.0.
Agilent Technologies, 2015b. λ Select-cII Mutation Detection System for Big Blue
Rodents INSTRUCTION MANUAL Catalog #720120 Revision B.0.
Aoki, Y., 2017. Evaluation of in vivo mutagenesis for assessing the health risk of air
pollutants. Gene Environ. 39, 16. https://doi.org/10.1186/s41021-016-0064-6.
Bercu, J.P., Zhang, Shaofei, Sobol, Zhanna, Escobar, Patricia A., Van, Phu, Schuler, Maik,
2023a. Comparison of the transgenic rodent mutation assay, error corrected next
generation duplex sequencing, and the alkaline comet assay to detect dose-related
mutations following exposure to N-nitrosodiethylamine. Mutat. Res. Genet. Toxicol.
Environ. Mutagen 891.
Bercu, J.P., Masuda-Herrera, M., Trejo-Martin, A., Sura, P., Jolly, R., Kenyon, M.,
Thomas, R., Ponting, D.J., Snodin, D., Tuschl, G., et al., 2023b. Acceptable intakes
(AIs) for 11 small molecule N-nitrosamines (NAs). Regul. Toxicol. Pharmacol. 142,
105415 https://doi.org/10.1016/j.yrtph.2023.105415.From.NLM.Publisher.
Bringezu, F., Simon, S., 2022. Salmonella typhimurium TA100 and TA1535 and E. coli
WP2 uvrA are highly sensitive to detect the mutagenicity of short Alkyl-N-

10

32 / 34 ページ