Nitrites, Scavengers and AI: What the FDA’s 2026 Workshop Revealed About Nitrosamine Control

By Philippe Tschopp | October 2026

How low should nitrite levels in pharmaceutical excipients be? Which scavengers actually work? Can manufacturing processes introduce additional nitrosamine risks? And can artificial intelligence help formulators predict problems before they arise?

These questions featured prominently at the FDA’s Fiscal Year 2026 Generic Drug Science and Research Initiatives Public Workshop, held on 8–9 June 2026.

Session 1 brought together researchers and experts from the FDA, Merck, Sandoz, Teva, Apotex, Eli Lilly, Gilead and academia to discuss the continuing challenges of nitrosamine impurities. A related presentation by BASF in Session 2 explored AI-assisted formulation and nitrosamine risk prediction.

The presentations provide useful experimental results and practical experience, particularly for formulation scientists and excipient manufacturers. They also highlight areas where scientific evidence and regulatory expectations are still developing.

Nitrite levels in excipients: What does the data show?

One of the most relevant contributions came from Ulrich Reichert of Merck KGaA (MilliporeSigma), who presented results from nitrite testing across more than 450 excipient products, generally evaluating two to three batches per product.

| Measured nitrite concentration | Share of tested products |
|—|—:|
| ≤ 0.05 µg/g (50 ppb) | 62% |
| ≤ 0.10 µg/g (100 ppb) | 67% |
| > 0.10 to 0.50 µg/g | 28% |
| > 0.50 µg/g (500 ppb) | 5% |

These results confirm that low nitrite concentrations are achievable in many pharmaceutical excipients. However, approximately one-third of the products tested contained more than 100 ppb.

Reichert also highlighted the influence of raw materials, manufacturing processes and excipient suppliers. Nitrite content can be affected by basic inorganic starting materials, heat treatment, spray drying and washing or purification steps.

The data support an important practical conclusion: an excipient’s chemical identity alone is insufficient to predict its nitrite content. Supplier-specific and batch-specific information remains essential.

There is no universal FDA limit of 50 or 100 ppb for nitrite in pharmaceutical excipients. These values describe the experimental distribution presented by Merck. Appropriate control levels depend on the specific formulation, nitrosation potential, dose and acceptable intake of the resulting nitrosamine.

Scavengers: Some impressive results, but no universal solution

The presentations from the FDA, Merck, Teva and Sandoz provided particularly interesting evidence on nitrosamine mitigation through formulation changes.

In a three-hour laboratory screening using bisoprolol as a model compound, Merck reported the following reductions in nitrosation:

| Scavenger | Reduction in the screening experiment |
|—|—:|
| Ascorbic acid | 100% |
| L-Cysteine | 100% |
| Glutathione | 100% |
| Caffeic acid | 99% |
| Ferulic acid | 97% |
| Tryptophan | 85% |
| Vitamin E | 55% |

These were short-term screening results under defined experimental conditions. They should not be interpreted as guaranteed performance in finished drug products.

Merck also observed tablet discoloration with certain scavengers, including ascorbic acid and tryptophan, during stressed stability testing.

The FDA’s presentation by Diaa Shakleya provided more detailed formulation data. In nitrite-spiked experimental tablets, ascorbic acid demonstrated a clear concentration-dependent inhibition of nitrosamine formation during six months of stability testing. At 40°C/75% relative humidity, formulations containing 0.1%, 0.5% and 1% ascorbic acid achieved reductions of approximately 71%, 86% and 92%, respectively. Caffeic acid and ferulic acid were less effective under the same experimental conditions.

The data demonstrate that both antioxidant identity and concentration matter, and that effectiveness can change with storage conditions.

Teva’s Valerie Niddam-Hildesheim added an important industrial perspective. She presented examples involving cysteine and propyl gallate, showing that different products containing the same nitrosamine may require very different remediation strategies. Even when the nitrite content of the excipients is controlled, the appropriate scavenger and concentration depend on the API, dosage form, formulation composition and manufacturing process.

Teva also highlighted practical disadvantages, including potential discoloration with ascorbic acid, odor problems with cysteine and limitations on the use of tocopherol in certain dosage forms.

pH modification can be surprisingly effective

Another important finding concerned modification of the formulation microenvironment.

In Merck’s experimental tablet studies, four pH modifiers were investigated at 5% concentration. After four weeks under accelerated conditions, the reported prevention of nitrosamine formation was:

| pH modifier | Prevention of nitrosamine formation |
|—|—:|
| Sodium bicarbonate | 100% |
| Sodium acetate | 100% |
| Disodium hydrogen phosphate | 31% |
| Trisodium citrate | 6% |

The results were formulation-specific, but the differences were substantial. The FDA’s own research also supported pH management as a mitigation measure. In its six-month experimental data, a sodium bicarbonate formulation achieved approximately 89–90% inhibition.

For formulators, local pH can therefore be an important variable alongside nitrite concentration and antioxidant selection. The consequences for API stability, dissolution and bioequivalence must nevertheless be assessed.

Sandoz: Nitrite control below 50 ppb in a practical case study

Filip De Bock of Sandoz presented two instructive case studies.

In the first, a secondary amine degradation impurity of the API reacted with nitrite originating primarily from microcrystalline cellulose and the API itself.

The formulation had a pH of approximately 5 and an acceptable nitrosamine intake of 100 ng/day.

Sandoz implemented several controls, including nitrite levels below 50 ppb in the MCC and API, additional control of the amine precursor and limits on nitrosamine impurities in the API. Reaction-network kinetic modelling and stability testing supported the strategy. Nitrosamine levels in the finished dosage form remained at or below 30% of the acceptable intake through shelf life.

This is a useful example of how targeted control of excipient impurities can prevent the need for a composition change.

In another case involving a prolonged-release formulation, a scavenger-based approach successfully reduced nitrosamine levels. However, the redevelopment project was eventually stopped because of costs and uncertainty surrounding regulatory requirements and bioequivalence.

The lesson is that technical success does not always translate into a commercially viable reformulation.

An additional source of nitrites: The air around the formulation

Jörg Schlingemann from Merck presented research on atmospheric nitrogen oxides (NOx) and nitrosamine formation.

The studies demonstrated that susceptible APIs can undergo nitrosation through exposure to atmospheric nitrogen oxides. Free-base forms were substantially more reactive than corresponding salts, with differences reaching several orders of magnitude.

The experiments also indicated that sample size and exposed surface area matter. Small samples exposed to air may accumulate nitrosamines more readily than larger bulk quantities. This is particularly relevant to fluid-bed processing, jet milling and other manufacturing operations involving considerable air contact.

A supplementary experiment involving different excipients showed substantial differences in their ability to convert and retain nitrite after exposure to NOx. This finding deserves attention because nitrite risk may change after excipient manufacture, depending on exposure conditions, processing and storage.

Schlingemann also discussed recovered solvents. Modelling and experimental work demonstrated that well-designed distillation processes can effectively remove many nitrosamines. The risk associated with recovered solvents therefore requires process-specific assessment rather than a blanket classification.

Measuring nitrosamines remains an analytical challenge

Houri Simonian from Apotex described the difficulties associated with developing analytical methods for nitrosamine drug substance-related impurities (NDSRIs).

Many NDSRIs require compound-specific LC-MS/MS or high-resolution mass spectrometry methods, together with appropriate reference standards and sufficient sensitivity.

One particularly relevant problem is artificial nitrosamine formation during sample preparation. Nitrite-containing samples exposed to acidic extraction conditions, elevated temperatures or prolonged preparation times can generate misleading results.

Matrix effects from excipients, co-eluting compounds and limited availability of analytical standards further complicate quantification. Reliable methods must therefore distinguish nitrosamines already present in the drug product from those potentially formed during analytical testing.

Risk assessment is evolving alongside formulation science

Several presentations addressed the toxicological assessment of nitrosamines and the development of scientifically justified acceptable intake limits.

The FDA’s Robert Dorsam discussed progress in the Carcinogenic Potency Categorization Approach (CPCA), enhanced Ames testing and quantitative in vivo mutagenicity assessment.

Research presented by Joel Bercu of Gilead and Robert Jolly of Eli Lilly indicated that in vivo mutation data may help predict relative carcinogenic potency. Jolly reported a strong correlation between transgenic rodent mutagenicity results and carcinogenic potency, with an R² of approximately 0.95 in the evaluated dataset.

Stephanie Simon of Merck presented an example involving N-nitroso-bisoprolol. Using error-corrected sequencing, benchmark dose modelling and an anchor compound, the researchers calculated a proposed acceptable intake of approximately 64 µg/day, considerably higher than the 1.5 µg/day structure-based value discussed in the presentation.

This is an investigational, compound-specific risk assessment result and should not be interpreted as a newly adopted regulatory limit.

The wider scientific discussion included less-than-lifetime exposure, uncertainty in potency prediction and the potential use of quantitative mutation data in future regulatory assessments. These areas remain important because a scientifically justified change in an acceptable intake could substantially alter the remediation required for a drug product.

BASF: Bringing AI into formulation risk assessment

One of the most forward-looking presentations came from Sandip Tiwari of BASF, titled AI-Assisted Generic Formulation and Nitrosamine Risk: Algorithmic Workflows.

BASF presented a digital workflow using its ZoomLab platform to support nitrosamine risk assessment during pharmaceutical formulation development.

The approach combines information on API chemical structure, predicted NDSRIs, acceptable intake limits, excipient composition, nitrite contribution and manufacturing scenarios. The workflow allows scientists to examine different excipient selections and compare processes such as direct compression and wet granulation.

For example, the model can estimate the contribution of individual excipients to the potential nitrosamine intake and identify which changes might offer the greatest reduction in risk.

This could make formulation screening and experimental prioritisation more efficient, particularly when several excipient combinations and manufacturing routes are being considered. However, the presentation also acknowledged important limitations, including uncertainty in nitrite conversion, supplier and lot variability, process-dependent formation kinetics and the need for harmonised datasets.

The examples demonstrated decision-support workflows. Broad validation against independent industrial datasets remains necessary before such predictions can replace experimental confirmation.

What does this mean for excipient manufacturers and formulators?

The most consistent message across the workshop was the need to combine excipient quality information with a better understanding of nitrosamine formation during drug product manufacture and storage.

The scientific evidence increasingly supports several complementary measures: characterised low-nitrite excipients, supplier and batch qualification, targeted use of scavengers, microenvironmental pH management and control of heat, moisture and atmospheric exposure.

The challenge is selecting the right combination for each formulation while maintaining its quality, stability, manufacturing performance and bioequivalence.

A further development is the growing use of predictive tools. Reaction kinetic modelling, statistical risk assessment and AI-assisted formulation workflows could help identify high-risk combinations before expensive development and stability studies begin.

For the excipient industry, the implications are clear. More transparent information on trace impurities, improved analytical capabilities and better understanding of functional performance under different processing conditions will become increasingly valuable.

The next advance in nitrosamine mitigation may come from combining better excipient data with predictive formulation science.

The FDA workshop offers a useful indication of how research is progressing, while also showing how much work remains before a fully predictive approach becomes routine.

 

Related FDA workshop presentations

All presentations below are available as original PDFs on the FDA website.

Session 1: Ongoing Challenges with Nitrosamines Impurities

1. Nitrosamines in Drugs: A Regulatory Perspective on Research Progress and Opportunities — Robert Dorsam, FDA. Regulatory research priorities, CPCA refinement, safety assessment and mitigation strategies.
2. Overview of HESI Program and Research Initiatives — Joel Bercu, Gilead Sciences. Enhanced Ames testing, less-than-lifetime exposure and collaborative nitrosamine toxicology research.
3. Using In Vivo Transgenic Rodent (TGR) Data to Evaluate Mutagenic Potency — Robert Jolly, Eli Lilly. Quantitative mutation data, potency correlations and potential methods for deriving compound-specific acceptable intakes.
4. Deriving Scientifically Sound Acceptable Intake Limits for N-Nitroso-Bisoprolol Using Error-Corrected Sequencing and Benchmark Dose Analysis — Stephanie Simon, Merck. A case study demonstrating how quantitative in vivo data could support a less conservative compound-specific intake calculation.
5. Nitrosamine Mitigation Strategies: Insights from the FDA and Emerging Challenges — Diaa Shakleya, FDA. Experimental evaluation of antioxidants, nitrite exposure, pH and storage conditions.
6. Antioxidants in Nitrosamine Risk Mitigation: Challenges and Considerations — Valerie Niddam-Hildesheim, Teva Pharmaceuticals. Industrial formulation case studies, antioxidant selection, regulatory challenges and remediation timelines.
7. Cracking the NDSRI Code: Mitigation in Generic Drugs, a Mission Impossible? — Filip De Bock, Sandoz. Practical case studies involving MCC nitrite control, scavengers, kinetic modelling and bioequivalence challenges.
8. Analytical Methods for NDSRIs: Challenges & Troubleshooting — Houri Simonian, Apotex. Analytical sensitivity, reference standards, matrix interference and avoidance of sample preparation artefacts.
9. Nitrosamine Risk of Recovered Solvents & NOx as Nitrosating Agent — Joerg Schlingemann, Merck. Nitrosamine removal during solvent recovery, atmospheric NOx and the impact of excipient exposure.
10. The Role of Excipients in Nitrosamine Risk Mitigation: Nitrite Control and Scavengers — Ulrich Reichert, Merck KGaA / MilliporeSigma. Nitrite measurements across more than 450 excipient products, scavenger screening and pH modifier studies.

Session 1: Public comment presentations

1. Other N-Nitroso Compounds — Raphael Nudelman, Nudelman ChemTox Consulting. Scientific gaps affecting other N-nitroso compound classes excluded from the conventional CPCA and proposed research to improve risk assessment.
2. Using In Vivo Mutation Data for Risk Assessment of Nitrosamine Impurities — George Johnson, Swansea University. The scientific rationale for quantitative mutation endpoints and their potential use in cancer risk assessment.
3. VICT3R Virtual Control Groups: Towards Qualification and Regulatory Acceptance — Frank Bringezu, EMD Serono. AI-supported virtual control groups designed to reduce animal use in toxicological research. Relevant to future safety assessment methodologies, although not specifically focused on excipient nitrites.

Session 2: Using AI to Address Practical Challenges for Generic Drugs

1. AI-Assisted Generic Formulation and Nitrosamine Risk: Algorithmic Workflows — Sandip Tiwari, BASF. AI-assisted NDSRI prediction, CPCA-based intake estimation, excipient contribution analysis and formulation/process scenario comparisons.

 

Additional regulatory references

FDA. Fiscal Year 2026 Generic Drug Science and Research Initiatives Public Workshop, 8–9 June 2026.

FDA. Control of Nitrosamine Impurities in Human Drugs, Revision 2, September 2024.

FDA. CDER Nitrosamine Impurity Acceptable Intake Limits, including updates published on 24 September 2026. Current compound-specific limits should be checked here rather than taken from historical workshop slides.

Editorial note: Experimental findings and industrial case studies summarised here represent the respective presenters’ work. Workshop contributions and proposed risk assessment methods do not themselves constitute new FDA guidance.