Nitrosamine impurities in peptides: where NDSRIs come from and why routine testing does not look for them

Most impurity classes discussed in connection with synthetic peptides are structurally close relatives of the target: deletion sequences, truncations, epimers, oxidation products, deamidation products. They arise from the synthesis or from the peptide’s own decomposition, and with enough chromatographic resolution they show up as peaks in the same analysis used to report purity. N-nitrosamines are different in almost every respect. They form after the peptide is finished, from a reaction between the peptide and something else in its environment; they are controlled at concentrations several orders of magnitude below the limit of a standard purity method; and they are regulated on a toxicological basis rather than a manufacturing-quality one. A certificate of analysis reporting 99.1% purity by RP-HPLC says nothing at all about them.

Since 2018, nitrosamine findings have driven recalls of more than a thousand drug product lots globally, and the regulatory apparatus built in response has expanded from a handful of small-molecule contaminants to a much broader category defined by structure rather than identity. Peptides sit inside that category more often than their chemistry might suggest. This post covers the formation chemistry, which peptide structural features are actually at risk, where the nitrosating agent comes from, how the current regulatory framework sets limits, and why detection requires a purpose-built method.

The nitrosation reaction and what it requires

N-nitrosation is the transfer of a nitroso group (–N=O) to an amine nitrogen. The reaction requires two things: an amine capable of forming a stable nitrosamine, and a nitrosating agent. Under aqueous acidic conditions the operative electrophile is usually dinitrogen trioxide (N₂O₃), formed from nitrous acid, which itself forms when nitrite is protonated. The pH dependence follows from this: the rate passes through a maximum around pH 3 to 4, where enough nitrite has been converted to nitrous acid but the amine has not been fully protonated into an unreactive ammonium form.

The amine’s substitution pattern determines whether a stable product results. Secondary amines give N-nitrosamines that are chemically stable and can persist through processing and storage. Primary amines are nitrosated readily but the product decomposes through a diazonium intermediate, so no stable nitrosamine accumulates — the lysine ε-amino group, despite being the most abundant free amine on many peptides, is not the concern. Tertiary amines can react but do so far more slowly, typically requiring dealkylation first. Studies indicate that in practice the secondary amine is rarely the limiting reagent; nitrite is present at trace levels while the amine is present at percent levels, so the amount of nitrosamine formed is generally governed by how much nitrosating agent the system contains.

This asymmetry has a practical consequence that is easy to miss. Because the peptide is in vast excess, controlling nitrosamine formation is almost entirely a matter of controlling nitrite, water activity, and pH — not of modifying the peptide.

Which peptide structures are actually susceptible

The peptide backbone itself is largely inert here. A backbone amide nitrogen is not a secondary amine in the reactive sense; conjugation with the carbonyl removes most of its nucleophilic character, and amides are poor nitrosation substrates under the conditions that matter. The risk therefore comes from specific structural features rather than from being a peptide.

The clearest case is proline. Proline’s nitrogen is a secondary amine held in a pyrrolidine ring, and when proline occupies the N-terminal position that nitrogen is free. N-terminal proline is common in research peptides, and hydroxyproline behaves similarly. Sequences carrying N-methylated residues present the same situation wherever the N-methyl amide is hydrolyzed or where the modified residue sits at the terminus. Synthetic analogs increasingly incorporate non-natural fragments — piperazines, piperidines, N-alkyl linkers, and the secondary amine functionality present in some conjugation chemistries — each of which introduces a nitrosatable site independent of the sequence.

A second and less obvious route is degradation-derived. A peptide with no free secondary amine as synthesized can generate one through hydrolysis at an N-alkylated position, exposing a nitrogen that was previously amidated. This means nitrosamine risk is not a fixed property of a sequence; it can increase over the material’s storage life in parallel with the degradation pathways covered elsewhere on this site.

Where a nitrosamine does form, the mass change is small and specific: replacing a hydrogen on the amine nitrogen with a nitroso group adds 28.990 Da. That is a tractable signature for high-resolution mass spectrometry, but only if the species is present at a concentration the instrument can see — which, as discussed below, is the crux of the problem.

Where the nitrite comes from

Nitrosating agents are not deliberately introduced. Nitrite has been characterized as a ubiquitous trace contaminant, and research into solid drug products has identified excipient-borne nitrite as the dominant source in most formation events. Reported levels in common pharmaceutical excipients span roughly the low parts-per-million range, which is far below any specification anyone was previously setting, yet more than sufficient to generate nitrosamine at the nanogram-per-day levels now being controlled.

For research peptides the relevant inventory is different from a tablet formulation but not shorter. Candidate contributors include water and diluents used for reconstitution, residual reagents and solvents carried through synthesis and purification, bulking agents and buffer salts present in the lyophilized cake, elastomeric closures and their cure packages, printed or adhesive packaging components in contact with the container, and the ambient atmosphere in facilities where nitrogen oxides are present. Nitrite content is not routinely reported for any of these.

Two secondary factors modulate the outcome. Residual moisture governs molecular mobility in the solid state, so the same considerations that determine hydrolysis and deamidation rates in a lyophilized cake also determine whether a nitrosation reaction can proceed at all. And local pH in the amorphous solid — the “pH memory” retained from the pre-lyophilization solution — determines whether the system sits near the nitrosation rate maximum or well away from it.

How the limits are set, and what that means for research-grade material

FDA’s framework distinguishes small-molecule nitrosamines from nitrosamine drug substance-related impurities (NDSRIs), which are nitrosamines formed by nitrosation of the active ingredient itself or a fragment of it. A nitrosated peptide falls squarely into the second category. The agency’s August 2023 final guidance recommends acceptable intake (AI) limits for NDSRIs, and a September 2024 revision of the broader control guidance sets out the risk-assessment, testing, and reporting expectations.

The AI limit is defined as an exposure approximating one additional cancer case per 100,000 people assuming daily exposure across a lifetime. Limits are assigned by a carcinogenic potency categorization approach that scores structural features around the nitrosamine nitrogen rather than requiring compound-specific carcinogenicity data for every new NDSRI. Where no limit can be derived by the available approaches, the guidance recommends a default of 26.5 ng/day.

That framing is worth reading carefully, because it is dose-anchored. An AI limit in nanograms per day converts to a concentration specification only after dividing by an assumed daily intake — and for material designated research use only, no such intake exists. The same structural argument applied to elemental impurities under ICH Q3D applies here: the toxicological limits describe a regulatory regime that RUO material does not sit inside, and quoting them as a concentration specification for research-grade peptide misrepresents what the number means. What survives the translation is the mechanistic and analytical content: which structures are susceptible, what conditions promote formation, and what it would take to look.

Why routine analysis does not see it

Consider the concentration scale. An impurity controlled at tens of nanograms per day against a milligram-scale substance sits in the parts-per-billion to low parts-per-million range relative to the peptide. A standard RP-HPLC purity method with UV detection at 214 nm has a practical reporting threshold around 0.05% — roughly 500 parts per million. The controlled level is two to four orders of magnitude below what the method can report. The nitrosamine is not being missed because it co-elutes or because the gradient is wrong; it is below the detection floor by a margin no amount of method optimization within that technique will close.

Intact-mass ESI-MS has the same problem from a different direction. The +28.990 Da satellite is resolvable in principle, but a species at parts-per-million relative abundance does not produce a usable ion signal against the target peptide’s isotope envelope in a routine identity run.

Detection requires a dedicated method: liquid chromatography with tandem mass spectrometry or high-resolution mass spectrometry, operating in a targeted mode, with a validated sample preparation and a reference standard for the specific nitrosated species. Method development is compound-specific and non-trivial, and one recurring pitfall is in-situ artifact formation — nitrosamine generated during sample preparation itself, from residual nitrite in the extraction solvent under the acidic conditions often used, producing a result that reports the analysis rather than the sample. Work on glycopeptide NDSRIs using high-resolution instrumentation has explicitly addressed distinguishing genuine impurity from preparation artifact.

Nitrosamine impurity control is an example of a broader pattern that recurs across peptide analysis: the routine testing panel answers the question it was designed to answer — is this the right molecule, and how much of the chromatographic area does it account for — and is structurally incapable of answering adjacent questions. Elemental impurities, disulfide connectivity, chiral purity, and residual moisture each fail against area-percent purity for their own reasons, and nitrosamines fail for a reason that is simply arithmetic.

For anyone evaluating a peptide’s impurity profile, the useful posture is not to demand nitrosamine data on an RUO certificate of analysis, where it will not appear and where the regulatory limits would not translate anyway. It is to know which structural features — an N-terminal proline, an N-methylated residue, a piperazine or other N-alkyl fragment — put a given sequence in the susceptible class, and to recognize that the same storage variables that govern every other degradation pathway, moisture and pH and time, govern this one as well.