How regulators keep cancer-causing traces out of your medicines

Trace impurities that carry no therapeutic purpose can hide inside even the most carefully manufactured medicines, and a new review published in the journal Genes and Environment lays out how the international regulatory system hunts them down. The article, authored by Masamitsu Honma and published on 23 December 2025, examines the International Council for Harmonisation (ICH) M7 guideline, the framework first issued in June 2014 for assessing and controlling DNA-reactive, mutagenic impurities in pharmaceuticals. According to the review, mutagenic impurities are of particular concern because even very low levels of exposure can theoretically increase the risk of cancer in humans, a property that sets them apart from ordinary impurities governed by conventional threshold-based rules.

The reason for this special treatment lies in the biology of mutation. The review explains that mutagenicity refers to a substance’s ability to induce mutations, typically through direct chemical interaction between the compound and DNA. Mutations are irreversible and permanent changes in the genome, and even a single mutation can potentially initiate carcinogenesis. For that reason, mutagenicity is generally regarded as a non-threshold phenomenon: there is assumed to be no exposure level that is entirely without risk. This stands in contrast to the traditional impurity guidelines ICH Q3A (R2) and ICH Q3B (R2), which establish qualification thresholds for impurities based on the maximum daily dose of a drug. Those thresholds, the review stresses, simply cannot be applied once an impurity is known or suspected to be mutagenic.

Since its initial release, the M7 guideline has evolved through a series of revisions. The first, ICH M7(R1) in March 2017, added an addendum defining acceptable intake (AI) levels and calculation methods for 14 mutagenic or carcinogenic compounds frequently detected in pharmaceutical manufacturing. The second, ICH M7(R2) in April 2023, added seven more compounds and introduced minor textual updates. In between, after several years of implementation revealed variations in interpretation among regulatory authorities, the ICH published a Questions and Answers document in May 2022 consisting of 25 individual questions and answers, designed to promote consistent understanding and global harmonization. Now a further revision, M7(R3), is in development under a maintenance work plan issued in February 2024, with a new topic on N-nitrosamine impurities formally adopted at the ICH meeting in Fukuoka in June 2024.

The scope of the guideline is narrower than many patients might assume. The review notes that ICH M7 applies mainly to new active substances and formulations during clinical development and at the time of marketing authorization applications. It does not generally apply to already marketed products, except where changes in the synthesis route, formulation, composition, or manufacturing process may affect carcinogenic risk, or where indications or routes of administration are modified. It also excludes drugs intended for the treatment of advanced cancer as defined in the scope of ICH S9, as well as flavoring agents, colorants, fragrances, and excipients previously used in approved products. However, a newly introduced, chemically synthesized excipient falls under M7, meaning the impurities it contains must be assessed.

At the heart of hazard assessment sits the bacterial reverse mutation test, better known as the Ames test. The review describes it as the essential assay for evaluating the mutagenic potential of pharmaceutical impurities, because mutagenic carcinogens of the DNA-reactive type are generally detected by this method. Compounds that test positive in chromosomal aberration or micronucleus assays but negative in the Ames test fall outside the scope of M7, since non-mutagenic genotoxic effects usually possess thresholds and do not pose a carcinogenic risk at impurity levels typically encountered in pharmaceuticals. When a positive Ames result may reflect bacterial-specific mechanisms irrelevant to humans, in vivo follow-up assays such as transgenic rodent gene mutation assays and Pig-a assays can clarify whether the response has biological significance in vivo. The review also highlights an emerging technique, error-corrected next-generation sequencing, which can detect induced mutations across the entire genome using DNA samples already collected from existing toxicology studies, reducing animal use.

When experimental testing is impractical, the guideline turns to computational chemistry. ICH M7 was, according to the review, the first international guideline to formally endorse the use of quantitative structure–activity relationship (QSAR) models as alternatives to biological assays for human health risk assessment. Impurities are often present only at trace levels, may consist of multiple chemical types, or are chemically unstable, making isolation or synthesis for testing difficult or impossible. M7 requires two complementary QSAR methods: expert rule-based models that identify structural alerts tied to mutagenicity, and statistical-based models built from large curated Ames test datasets. If both systems indicate no structural alerts, the impurity is deemed to have no mutagenic concern. The review notes that when M7 was established in 2014, reported sensitivities of QSAR tools for novel chemical entities were below 50 percent, but following improvement efforts including the international Ames/QSAR Challenge Project, average predictive accuracy has risen to approximately 85 percent.

For quantifying risk, the guideline leans on the threshold of toxicological concern, or TTC. For mutagenic substances lacking carcinogenicity data, ICH M7 sets the TTC at 1.5 micrograms per day, a value derived from an analysis of carcinogenicity data on more than 1,000 chemicals. The analysis estimated that, assuming roughly 10 percent of all chemicals are carcinogenic, a daily exposure of 1.5 micrograms would ensure a lifetime excess cancer risk of less than 1 in 100,000 for 99 percent of chemicals. Certain highly carcinogenic classes, known as Cohorts of Concern, are excluded from this general value: aflatoxin-like compounds, N-nitroso compounds, and alkylazoxy compounds. Where rodent carcinogenicity data do exist, a compound-specific AI is derived from the TD50, the chronic dose inducing tumors in 50 percent of test animals, using linear extrapolation to a lifetime excess cancer risk of 1 in 100,000. The review also describes less-than-lifetime adjustments based on Haber’s rule, which allow higher daily limits for shorter treatment durations, and a cumulative cap when three or more mutagenic impurities are present.

The framework’s most severe real-world test came in July 2018, when the antihypertensive drug valsartan, manufactured by a Chinese API supplier, was found to contain N-nitrosodimethylamine (NDMA) and N-nitrosodiethylamine (NDEA), both highly potent mutagenic carcinogens belonging to the Cohorts of Concern. The discovery triggered a global recall, and N-nitrosamines were subsequently detected in other medicines including ranitidine and metformin, prompting further recalls worldwide. The compound-specific AI values derived from rodent data are 0.0959 micrograms per day for NDMA and 0.0265 micrograms per day for NDEA, several orders of magnitude below the generic TTC. Since 2019, additional low-molecular-weight N-nitrosamines and drug-substance-related N-nitrosamines formed by nitrosation of the parent drug have been reported. For the latter, the review describes the Carcinogenic Potency Classification Approach, a provisional method that scores structural features influencing metabolic activation or deactivation and integrates existing carcinogenicity data, now formally accepted by EMA, FDA, and Japan’s MHLW for estimating AI values pending compound-specific data.

The review closes with a reflective note that reaches beyond regulatory procedure. In October 2024, Dr. Bruce N. Ames, who developed the Ames test in the 1970s, passed away. From the 1980s onward, the review recounts, Ames argued that the carcinogenic potential of synthetic chemicals had been overestimated, contending that the main causes of cancer lie in natural dietary carcinogens, lifestyle factors, smoking, and high-fat diets rather than trace synthetic pollutants. The article points out that NDMA and related N-nitrosamines are ubiquitous environmental carcinogens found in grilled meat, bacon, beer, and other foods, and that NDMA can form endogenously in the human stomach through reactions between dietary secondary amines and nitrite. Total exposure from such dietary and physiological sources is estimated to exceed the AI limits established for pharmaceutical impurities. While stringent limits undoubtedly promote higher pharmaceutical quality, the author argues, a balanced, human-centered approach remains essential from a regulatory science perspective, harmonizing risk management with societal benefit as the M7(R3) revision takes shape.

Subject of Research: Regulatory assessment and control of mutagenic impurities in pharmaceuticals under the ICH M7 guideline

Article Title: Guidelines for the assessment and control of mutagenic impurities in pharmaceuticals

Article References: Honma, M. (2025). Guidelines for the assessment and control of mutagenic impurities in pharmaceuticals. Genes and Environment, 47(1), Article 26. https://doi.org/10.1186/s41021-025-00349-5

Image Credits: AI Generated

DOI: 10.1186/s41021-025-00349-5

Keywords: ICH M7, mutagenic impurities, pharmaceuticals, Ames test, QSAR, threshold of toxicological concern, N-nitrosamines, NDMA, acceptable intake, drug safety, carcinogenic risk, regulatory science