Deriving safe limits for N-nitroso-bisoprolol by error-corrected next-generation sequencing (ecNGS) and benchmark dose (BMD) analysis, integrated with QM modeling and CYP-docking analysis

Abstract

N-nitroso-bisoprolol (NBP) is a nitrosamine drug substance-related impurity (NDSRI) of bisoprolol, which is used to treat cardiac diseases since decades. To investigate the mutagenic potential of NBP, in vitro methods such as Enhanced Ames Test (EAT) and a mammalian cell gene mutation (HPRT) assay were used. To assess the in vivo mutagenicity, a 28-day repeat-dose study was conducted in wild-type NMRI mice, and liver and bone marrow samples were subjected to error-corrected next-generation sequencing (i.e., duplex sequencing) followed by benchmark dose analysis (BMD). NBP did not show mutagenic effects in Ames tests using 10 % and 30 % induced rat or 30 % uninduced hamster S9. However, relevant increases in mutation frequencies were observed in an EAT in the presence of 30 % induced hamster S9 in strains TA100 and TA1535, confirming that the most stringent conditions of the EAT are appropriate to detect the mutagenic activity of weak mutagens, such as NBP. In the HPRT assay conducted in V79 cells, nitroso-diethylamine (NDEA) relevantly induced the mutation frequency, but not NBP. The highly sensitive error-corrected Next-Generation Sequencing (ecNGS) method to detect mutations across the genome represents an appropriate in vivo mutagenicity investigation equally suitable as a TGR assay to assess the mutagenic potential of nitrosamines. A weak induction of mutation frequencies was detected by ecNGS in the liver and the bone marrow of mice. Using BMD analysis, new safe limits were calculated for NBP, which are higher than the published AI of 1.5 µg/person/day. Using the approach to calculate Permissible Daily Exposure (PDE) limits according to ICH Q3C, a lifetime PDE of 400 µg/person/day was derived. Based on the ICH M7 framework for derivation of Acceptable Intake (AI) limits, an AI of 64 µg/person/day was established. Consistent with regulatory emphasis on mechanistic interpretation, in vivo modeling was further supported by in silico calculations. Specifically, the validated Computer-Aided Discovery and RE-design (CADRE) tool was used to predict the potency of NBP and further differentiate its metabolic activity from the anchor nitrosamine NDEA via quantum mechanics (QM) calculations and CYP-binding predictions. Outcomes of this analysis were consistent with in vivo studies, while offering a deeper understanding of the fundamental biochemistry using a physics-led method. The integrated in vivo–in silico investigation provides a data-based determination of safe limits, suggesting that the AI based on structural considerations solely might be over-conservative and should not be capped at the TTC.

Highlights

First presentation of in vivo mutagenicity data for an NDSRI of a beta-blocker

Nitroso-bisoprolol was not mutagenic in standard and modified Ames tests using rat or uninduced hamster S9, but positive in EAT with 30 % induced hamster S9

ecNGS revealed a low mutagenic potential, suggesting that nitroso-bisoprolol is not in the cohort-of-concern

Using benchmark dose analysis, new safe limits far above the published 1.5 µg/day could be derived, suggesting that the CPCA based on SAR considerations solely is over-sensitive and should not be capped at the TTC.

The new safe limits for one beta-blocker determined in the present work may serve as blueprint for class-specific PDE or AI to be applied to NDSRIs that bear an isopropyl or tert-butyl group connected to the nitroso group.

Horizontally integrated in silico–in vivo analysis shows good agreement between CADRE (QM) outcomes, augmented here with CYP-docking analysis, and in vivo results

In silico (QM) models can be used a priori to gauge feasibility of a higher AI and/or post in vivo studies to gain greater confidence in the proposed limit based on mechanistic interpretation

Introduction

N-nitrosamines belong to a class of N-nitroso compounds, which are referred to as members of the cohort-of-concern carcinogens according to the ICH M7 guideline, as some of them of are known potent rodent carcinogens and thus potential human carcinogens (ICH 2023). N-nitrosamine contaminations have been a point of concern for pharmaceutical manufacturers and regulators alike since their first discovery in batches of the angiotensin II receptor antagonist valsartan in late 2018 (Nudelman et al. 2023). While first reports concerned the small and potent dialkyl nitrosamines, such as nitroso-dimethylamine (NDMA) and nitroso-dieethylamine (NDEA), it was soon discovered that there is a second dimension related to N-nitrosamines derived from vulnerable Active Pharmaceutical Ingredients (APIs) and impurities, especially those that are secondary amines. Due to their structural similarity to the drug or fragments of the drug, those N-nitrosamines are referred to as “nitrosamine drug substance-related impurities” (NDSRIs) (EMA 2024a; FDA 2024). It was shown that this concerns a considerable percentage of available medicines, putting whole classes of drugs at risk of forming N-nitrosamines (Schlingemann et al. 2023). One of these classes are agents targeting the adrenergic beta receptors, which comprises both respective agonists and antagonists, the latter also being known as beta-blockers. An overview on common beta agonists and antagonists is provided as supplementary information.

Bisoprolol is a beta1-selective-adrenoceptor blocking agent (β-blocker) used in the treatment of cardiovascular diseases, such as hypertension, angina pectoris, and heart failure. It is included in the WHO list of essential medicines (WHO 2023). All β-blockers and β-agonists share a common structural motif that includes a hydroxyl group and a secondary amine functionality. For this reason, the whole class of compounds is at risk of forming N-nitrosamines under promoting conditions, i.e., in the presence of nitrosating agents. Nitroso-bisoprolol (NBP) is an NDSRI of bisoprolol. The chemical structures of bisoprolol and NBP are shown in Fig. 1.

**Fig. 1** Chemical structures of Bisoprolol (left) and Nitroso-bisoprolol (NBP) (right)

Regulatory agencies, such as EMA and U.S. FDA, have set provisional acceptable daily intake levels for a growing number of nitrosamines either calculated from TD50 values of lifetime rodent carcinogenicity studies, determined by extrapolation from close analogues (read-across), derived by the Carcinogenicity Potency Categorization Approach (CPCA) or based on Enhanced Ames Test (EAT) and/or in vivo mutagenicity data (EMA 2024b; FDA 2023; Kruhlak et al. 2024). The acceptable intake (AI) for NBP and most other NDSRIs from this class of medicines has been set to 1500 ng/day (EMA 2024b). While carcinogenicity data are available for many of the low-molecular-weight nitrosamines such as NDMA or NDEA, these data are and will not be available for most of the NDSRIs. This lack of data leads to uncertainties how to derive realistic and safe AI levels for NDSRIs across industry and regulatory agencies. The CPCA first introduced by Health Authorities (HA) in 2023 was a major step forward in the assessment and control of nitrosamines as it allowed a rapid and consistent assignment of provisional AIs to a large number of NDSRIs in the absence of any compound-specific data and reduced some uncertainties significantly (Bercu et al. 2024; EMA 2023; FDA 2024; Ponting et al. 2024). However, the perception of the CPCA differs between HAs. While it is generally accepted that the CPCA is a conservative approach, EMA considers all four options for nitrosamines without substance-specific data as described above equally suitable, whereas FDA seems to prefer the CPCA AI as the default (Kruhlak et al. 2024).

The bacterial reverse mutation (Ames) test is commonly accepted to assess the mutagenic potential of chemicals, including impurities (ICH 2023). To address concerns from HAs that the Ames standard protocols according to OECD 471 (OECD 2020) might not be sensitive enough to detect the mutagenicity of nitrosamines, considerable work has been conducted by industry and within agencies. As a first result, the Enhanced Ames Test (EAT) protocol was published by EMA in 2023, other health authorities followed that approach (Canada 2024; EMA 2024c; FDA 2023). The recommended conditions include the use of the pre-incubation method (30 min pre-incubation time), the application of 30 % induced rat as well as 30 % induced hamster S9, the inclusion of two additional nitrosamine controls, e.g., NDMA or CPNP and the use of the lowest volume of organic solvent as possible (EMA 2024c). A large ring trial was initiated by the Health and Environmental Science Institute (HESI) Genetic Toxicology Technical Committee (GTTC) nitrosamine subgroup involving more than 20 companies, institutions, and agencies to investigate the “Concordance between Ames and Rodent Carcinogenicity Outcomes for N-Nitrosamines (NAs) with Rat and Hamster Metabolic Conditions” and to identify the most sensitive Ames conditions. The results of this ring trial were presented on a joint FDA-CDER/HESI meeting and recently published by the working group (Bercu et al. 2025). In this ring trial, 30 % induced hamster S9 showed the highest sensitivity. It was agreed that both E. coli WP2uvrA strains (with or without plasmid) were equally suitable. Also, DMSO was considered an appropriate solvent (Bercu et al. 2025). There is still debate on whether nitrosamine positive controls should be included and if, which ones to use. To investigate suitable solvents and volumes further, an initiative has been started by the Lhasa Complex Nitrosamine Consortium in collaboration with industry members. In the light of the acceptability of negative EAT results, there are also differing requirements by HAs. EMA and other agencies allow control of a nitrosamine tested negative in an EAT at 1.5 µg/day (Canada 2024; EMA 2024a). The FDA guidance remains less specific on this point and the agency may request additional data to support a 1.5 µg/day limit (FDA 2023). Until now, it is not specified which data might be requested, however, e.g., in vitro mammalian cell mutagenicity or in vitro metabolism data have been discussed. These discrepancies between HAs are still challenging for industry as there is uncertainty on which data may be ultimately requested.

To achieve limits above 1.5 µg/day, EMA recommends using either a read-across approach from a suitable surrogate molecule with carcinogenicity data or data from a “relevant, well conducted in vivo mutagenicity study” and allows control of a nitrosamine to ICH Q3A/B levels, if the result is negative (EMA 2024a). As the type of study is not specified further, there is ongoing debate on which of the available in vivo mutagenicity assays is acceptable for HAs. For many years, in vivo mutagenicity assays using transgenic rodents such as Big Blue® or MutaMouse® were the assays of choice and are widely accepted by authorities. However, in the light of the nitrosamine topic, sequencing methods such as error-corrected Next-Generation Sequencing [i.e., Duplex Sequencing (DS)] evolved considerably and gained more and more importance as an in vivo mutagenicity method to detect even rare mutations (Marchetti et al. 2023a, 2023b; Salk and Kennedy 2020; Valentine et al. 2020). The properties of TGR and DS are summarized in Table 1.

Download the full article as PDF here Deriving safe limits for N-nitroso-bisoprolol by error-corrected next-generation sequencing (ecNGS) and benchmark dose (BMD) analysis, integrated with QM modeling and CYP-docking analysis

or read it here

Table 1. Properties of TGR assay and ecNGS

	TGR assay	ecNGS
Scope of mutation detection	Reporter gene-specific mutations (e.g., lacZ, gpt)	Genome-wide mutation detection, covering point mutations, insertions, deletions
Throughput	Low	High
Regulatory acceptance	Widely accepted, validated	Not fully validated yet for regulatory use
Test system/ethical considerations	Transgenic rodents	Wild type rodents (or any other species)
Costs	High (breeding and maintenance of transgenic animals, labor-intensive)	Medium (sequencing costs, computational analysis, potential to reduce in cost with time)
Time	Long lead time, globally limited breeding and testing capacities	Shorter lead time, sequencing capacities are currently growing
Data interpretation	Simple, based on a defined number of mutations	Complex, large datasets requiring advanced bioinformatics
Integrable in standard toxicity studies (3Rs)	No	Yes

Simon, S., Schlingemann, J., Johnson, G. et al. Deriving safe limits for N-nitroso-bisoprolol by error-corrected next-generation sequencing (ecNGS) and benchmark dose (BMD) analysis, integrated with QM modeling and CYP-docking analysis. Arch Toxicol (2025). https://doi.org/10.1007/s00204-025-04103-2