Could gelatin contains amine functionalities, which react with nitrosating agents to form nirosamines?

We were recently asked by a regulatory agency about this possibility in a capsule for which an azo dye is sued and therefore traces of nitrites are present.

1 Like

I was asked a similar question at a workshop here in US. not familiar with the particular case

For that case you should check the azo colorant synthesis. As of course is an azo coupling reaction, a nitrite salt is commonly used. What is the other "couple used, does it have a secondary amine moiety? As from what I known, not a single approved colorant for food have this characteristic. (Note apart, for coloured foils, there are some cases of colorants with amine moiety, but is a different topic with different answers).

What should also be considered is that azo colorants, and even in capsules, they are used in a very low %. If any unidentified secondary or tertiary amine is nitrosated, I would expect it to be below the 18 ng/day threshold. (I asked similar a question in a FDA workshop back in early 2021 and similar response was answered around “how much colorant do you use”).

Regarding the nitrite, I have also seen some manufacturers to use sulfamic acid as a nitrite scavenger. If there is no scavenger I would expect quite a bit of nitrite per mg of colorant at the end.

Also other points worth mention:

  • The colorants in question are normally expected to comply with FDA 21 CFR regulation for colorants subject to certification and (EU) 231/2012 regulation for food additives. Therefore, an expected acceptance criteria of purity and content of impurities is known. The “residue” after all known deductions is quite low.

  • Having said this, common impurities depending on the azo colorant are primary amines and triazene like compounds. None, would form an stable nitrosamine I would say.

The same would apply to nitrite, as the quantities of colorant per capsule are low, even lower quantities of unknown impurities will be present. Therefore, I expect more nitrite could be present in other common used excipients that the capsule itself (because of azo colorants).

Now, around the gelatin capsule itself and if there is any amine functionality. Collagen have amino acids. But most amino acids are primary amines, I think there are only a few that aren´t (proline, etc.) and are already known that if nitrosated are non carcinogenic. Once again, capsule gelatin composition should be known, but other common compounds part of its composition are water and mineral salts, so nothing strange af first glance.

Finally, just some clarifications, this “statement” works best when the capsules are not so “big” as well as the used quantity of azo colorant are not in a high % and its known to comply with FDA or EU regulation for colorants in food. Its a case by case, with some math behind depending of the maximun daily dose of the drug.


Many thanks all of you for the responses. I agree that the amount is critical and we can use that approach to justify the potential compound will be below the 18ng limit.

Also, pH of formulation is controlled around 7,0 from which the possibility of nitrite convert into a nitrosating agent is low. Do you agree this approach?

1 Like

I added the description of EFPIA DP workflow ver2.0.

“Capsule shells are also considered to be no identified risk for nitrosamine contamination through either colours used within the capsule coating, neutrality of the plasticisers used, or the external printing ink that may be used. Nitrosation is unlikely to occur from nitrite within the gelatin because pH is likely to be neutral, gelatin contains primary amine scavengers, the low surface to volume ratio and the printing process is conducted at room temperature.”

As @Diego_HM explained, the risk of nitrite contamination in gelatin doesn’t seem serious in many cases.


thanks for Wrokflows article

Have any regulatory authorities accepted these justifications for the risks associated with gelatin being low due to the lack of carcinogenicity?

Hi Diego,

Thank you for sharing your Knowledge regarding this subject. Would you have any reference to share about the carcinogenicity of amino acids that you mentioned?

I would also be interested in this subject.

1 Like

Has anybody trying NAP test on the empty capsule to assess potential formation?

Hi @brunacasanova,

Regarding your question, primary amines after nitrosation forms unstable compounds, therefore, its carcinogenicity is irrelevant if their mean life is short enough that it even do not reach the tissue of interest (e.g: liver).

For the aminoacids that have scondary amine groups or reasemble one like proline & hydroxyproline. Carcinogenic studies (a bit old nonetheless) indicates that are non-carcinogenic.

Nitroso-proline & Nitroso-hydroxypropline: Studies of the tumorigenic effect in feeding of nitrosamino acids and of low doses of amines and nitrite to rats - PubMed (

Nitroso-tryptophan is +tive in a bacterial test Mutagenicity of nitrosated α-amino acid derivatives N-acetyl-N’ -nitrosotryptophan and its methyl ester in bacteria | Carcinogenesis | Oxford Academic ( & Mechanisms and kinetics of tryptophan N-nitrosation in a gastro-intestinal model - ScienceDirect, but gelatin do not have tryptophan as part of its chains.


Thank you very much for all the information shared :+1:

In the context of the above, I would like to know if anyone has been able to receive information from the supplier about the structure of the biopolymer that constitutes the capsule? It may help to know the type of source from which the raw material is extracted, considering that, depending on its nature, the composition and properties of gelatin tend to vary to some extent.


1 Like

I would recommend to stay focused on the N-terminus and nitrosamines sensu stricto, while using nitrosamine calculation systems and leveraging data from the food context (besides applying the toxicity data on the related nitrosamines in the risk assessment). I don’t think it is so critical to have very accurate data on the gelatin composition of the source used to derisk.

The stability of the protein used (against proteolysis) and the low nitrite exposure and non-stringent conditions linked to gelatin use indeed truly limit the related nitrosoproline risks.

The general regulators recognition of the non-carcinogenicity of (hydroxy)proline would probably end the nitrosamine discussions on gelatin.

(Note that nitrosoproline is also measured as a biomarker to study endogenous nitrosation in humans.)

I’m aware of some international regulators who have endorsed detailed risk assessments to support no nitrosamine risks for gelatins though.

Often based on their structure and processing, and endorsing the absence of tryptophan if you want to dive into the nitroso in general:

Nitroso(hydroxy)proline is the only “real nitrosamine” one can think of in this context with a protein (and gelatin) and (hydroxy)proline does have significant relevance in the gelatin composition (where is present as an amide though).

A worst case estimation can be 20% proline and 15% hydroxyproline in gelatine.

Indeed, you can evaluate:

  • A: Proline as free amino acid that gets nitrosated (calculation needed to understand worst case) (can be tested without proteolysis)
  • B: Proline as amide in the chain (no risk because tertiary amide and assumption that the protein is stable (no proteolysis releasing new proline N-termini or free proline, cf. scenario C)
  • C: Proline at the N-terminus that gets nitrosated (calculation needed to understand worst case (based on nitrite, MW peptide and estimation of %proline in peptide and % thereof at N-terminus)) (cannot be tested without proteolysis in vivo or in the lab)

I recommend to evaluate or estimate the free amino acid content (which should be quite low) of the gelatin used, assume a significant part of that is proline/hydroxyproline (see literature on their prevalence in gelatins) and then calculate what this would mean intake-wise if everything was max. nitrosated (nitrite depletion if limiting reagent and standard reactivity assumptions for non-stoichiometric chemistry), while regards are being made to non-carcinogenicity of nitrosoprolines.
You can also assume at the end of the chain a certain prevalence of (hydroxy)proline and calculate that as worst case nitrosated with a certain prevalence.
And use literature data on nitrosoproline testing in food to underbuild certain assumptions.
For example data for meat shows that A+C can corresponds to 123 ng nitrosoproline per g meat, whereas A only corresponds to 20 ng nitrosoproline per g meat. (For some special meats sometimes levels of 4000 ng/g meat.
Already based on the meat data, a rough estimate of A+C for gelatine of 300 ppb shows no identified risk (as > 5 g gelatine/day via medicines is unlikely, whereas nitrosoproline is non-carcinogenic and should be allowed at much higher levels than 1500 ng/day and in reality your meat, beer,… consumption can lead to dozen times more exposure!)
(Dunn, B. P., & Stich, H. F. (1984). Determination of free and protein-bound N -nitrosoproline in nitrate-cured meat products. Food and Chemical Toxicology, 22(8), 609–613.)

If you are really interested in nitrosamines sensu lato, you can do the same for nitrosation of moieties like lysine, tryptophan and histidine (not for (hydroxy)proline as the build-in in the peptide chain protects from nitrosation).
For the gelatin case, just calculating lysine and histidine should be enough (lysine ca. 6% of gelatin, histidine <1% and tryptophan absent so no need to calculate this one).
As a peptide will eventually be hydrolysed in vivo (gelatin capsule is stable under processing and storage typically though, no protein cleavage expected), calculating with the molecular weights of the moiety and the fraction of the moiety is justifiable.
Your estimation will have to evaluate kinetic and steric factors as well when predicting the nitrosation degree (think like accessibility of the moieties based on the 3D structure of the protein).
For ε-nitrosolysine the calculation truly theoretic and exemplar, based on the reversibility of nitrosamine of primary amines, which Kuo et. al. 2004 have also suggested for nitrosolysine.

Reactivity calculations compared to acceptable intake estimations (and testing data from the food industry) and low nitrite in gelatin (but pending excipients/colorants) should be enough to derisk.

Note that also in the food context (cf. proteins), the European authorities prioritize nitrite exposure limitation as primary focus and the EFSA nitrosamine risk assessment report did contain little specific concerns directly related to nitrosoproline.

In my opinion there is a little value in NAP or ATNC testing on gelatin, as it will not be clear if the positive response is coming from nitrosamides, nitrosamines, nitrosated heteroaromatics, C-nitrosated product and other sorts of pseudonitrosamines (depending on the method sometimes even nitrite is just detected as a false positive). Of course a false positive will have to be proven to be just (the A from apparent might not be enough to ignore it) that and more specific testing on complex peptide substrates can be quite challenging (need for proteolysis, literature data on non-robustness (artefact generation)).

For an example where the ATNC does not accurately predict the nitrosoproline content: A method for the investigation of free and protein‐bound N‐nitrosoproline in beer: Food Additives & Contaminants: Vol 5, No 2 (

Of note, as nitrosoproline risks are typically higher for the N-terminus than for the free amino acids (based on meat testing), a hydrolysis of the peptides in the sample (enzymatic or chemically) will be required for sample prepping if you really want to test, but this can increase complexity and robustness issues.
If you are not planning proteolysis during testing of the capsule shell, you might as well not test.
An additional irony is that this proteolysis happens best in vivo for representative results. For foods one could better estimate the nitrosoproline content in meat by having volunteers eat the meat and evaluate nitrosoproline in urine, rather than digesting and testing meat samples in the lab: definitely meat products containing nitrite can get nitrosated during the proteolysis (release a free proline and then nitrosation), meaning the literature data is sufficiently worst case.

But I truly believe the meat industry has evaluated nitrosoproline in meat sufficiently, so that together with the proof of non-carcinogenicity we can fully address this for medicines in step 1 rather than step 2 where gelatin is being used. Gelatin suppliers should be quite supportive to assess this as such.

If you want to evaluate other types of stress (cf. pH) the nitrosoproline testing related to beer products can also be of interest.

In gelatin materials for pharma identifying a risk for small nitrosamines linked to certain colorants and other materials used during processing is possible and might require small nitrosamine testing to derisk. However, the other side of the spectrum where nitroso compounds linked to peptides are evaluated is my opinion unlikely to find connection with the call for review on nitrosamines.

  • Non-synthetic peptides have another risk profile for free amino acids nitrosated and being nitrosated than synthetic peptides; moreover the only true nitrosamine from a free amino acid would be N-nitroso-(hydroxy)proline. As nitrosoproline was found as not carcinogenic in a number of studies (assigned to the COOH potency reducing effect) and as significant amount of nitroso(hydroxy)proline as free amino acid in gelatin are unlikely, this possesses no risk.

  • For the N-terminus risks linked to hydroxyproline: evaluate the prevalence of this scenario (how often a proline as N-terminus?) and the testing data from food to evaluate the risks and the need for testing.

  • When (hydroxy)proline is in the middle of the chain: protected as amide no risk as long as no risk is identified for amide bond breakage (unlike during shelf life of medicines often, cf. resonance stabilisation of amide bond (stable)).

Primary amine related risks can be derisked as usual (see above on lysine).

For non-nitrosamine linked nitrosation (secondary amide bond in the chain (non-proline), tryptophan, histidine etc):

  • If you are interested in nitrosopeptides (nitrosamides): several Ames tests have been reported in literature on small (di/tripeptide) with such N-nitrosation (e.g. D. Anderson papers from the ’80).
  • N-nitroso amides and N-nitroso aromatics (N-nitroso tryptophan, N-nitroso histidine) are considered outside of the call for review though.
  • On nitrosated indoles (cf. tryptophan) several discussions on this forum exist.

Hi @ccdw thank you very much for the comments associated with the question I shared with the community. My consultation is based on the need, from a pharmaceutical perspective, to have as much detail as possible on the structural characteristics of all the components of the medicine and at the same time respond to regulatory requirements.

I find your ideas interesting, in this sense it would be ideal to be able to extend the analysis to the rest of the amino acids that make up gelatin. I share this research with you: Farris, S., Song, J., & Huang, Q. (2009). Alternative reaction mechanism for the crosslinking of gelatin with glutaraldehyde. Journal of Agricultural and food chemistry, 58, 998-1003, where they have reported, among others, a significant percentage of the following amino acids: Glycine, Alanine, Arginine and Serine.

I believe that it would be important to consider in the risk assessment, in addition to the above, the conditions of manufacture, storage and transport of the drug product, in this sense the effects of temperature, light and humidity at their different levels (including accelerated stability studies) become important, which will favor, among other transformations, the phenomenon of cross-linking making the initial matrix more complex. At this point, I believe that the NAP Test is an ideal candidate to demonstrate the formation or not of nitrosamines. To minimize the possibility of obtaining false positives, it would be advisable to make use of an analytical technique with adequate levels of selectivity and specificity, for these cases and for the required sensitivity, ideally mass detectors coupled to the chromatographic system.

1 Like

Hi all, maybe my two cents here.

To be very pragmatic here, is there a meaninful secondary amine as part of the gelatin source you are using? Maybe a cross linked, proline? We cant say for sure. Therefore, of most importance, is there enough nitrosating agents source to be able to react in the gelatin matrix and generate a meaninful amount of a nitroso compound? Also considering the possible steric hindrance both ex-vivo and in-vivo.

Going more into detail, competitive reactions will be in place. I would expect a high probability of the nitrosating agent to be depleted as also it may react with the primary amines that are in much percentage that any possible secondary amines even considering a cross linking. Actually, gelatin may be kind of a complex nitrosating agents scavenger. There is kind of enough evidence that gelatin do not have tryptophan. If, an experimental confirmation is needed, then a NAP test plus further analysis by an enough sensitive method for N-nitroso-tryptophan or a total nitrosamines analyzer could help. Nonetheless, in reality, the conversion factor will be much less with actual manuf. conditions.

Note: Tryptophan has an indole ring that is not per se a secondary amine. For propline and hydroxyproline, my comments are earlier into the post.

Now, about the cross linking, it is more favoured under an alkaline pH, where in exchange nitrosation is heavily related to an acid medium, unless a catalizer like an aldehyde is present. As indicated in the paper you shared “Therefore, at high pH values, very few amino groups become protonated, and there exists a large amount of free amino groups in gelatin molecules.”

Although, it is clear a robust risk assessment is the way forward, I would not lose my mind around the risk of gelatin if there are some gaps of information around the composition of my material, as this of course could vary. There is a non-negligible amount of information, both theoretical and experimental to fill the gap and dismiss the risk as long as the gelatin supplier is under a robust supplier qualification system. We want to avoid any kind of cross contamination or similar topics :slight_smile:


Dear All

Thank you very much for the time spent on this discussion and for the comments shared. All the aspects that were derived and that are possible to develop around the question I originally asked have been very interesting. There is no doubt that the aspects shared will allow me to strengthen the rationale used within the risk analysis.

Greetings to the entire community!

1 Like

I believe literature has shed enough light that this should be narrowed down to N-terminal (hydroxy)proline. Whether this yields enough nitrosamine depends on the nitrosation risks. New testing is probably needed to re-evaluate that most of the claimed detections of N-nitrosated N-terminal (hydroxy)proline in meat is artefactal (generated during sample preparation after hydrolysing protein to free proline building blocks, which could possibly also happen in vivo and has not elevated concern) and thus proof that the 3D structure of gelatin does not allow N-terminal nitrosation. The risk for nitrosation after hydrolysis during a nitrosation test (whereas the protein is maybe known to be stable from cradle to grave industrial use) will also have to be considered.

Nonetheless, gelatin has a high amount of proline and hydroxyproline, this is a sure as the absence of tryptophan.
Literature experiments on meat (as source of proline-rich proteins, a small part would be collagen although 3D structurally different from gelatin, and the precursor collagen would be amongst the most abundant proteins) have shown that the proteins in there are could be at risk for nitrosoprolines at the N-terminus of the chains. Although a partial nitrosation during sample preparation could not be excluded (as you have to hydrolyse the protein to release and test N-terminal nitroso(hydroxy)proline), probably this risk is not enough to fully claim absence of N-terminal nitrosation potential due to crosslinking effects. Interesting though if the crosslinking on gelatin doesn’t allow N-terminal nitrosation, then this could be more generally true. More complex structures like coils and helices (collagen) probably have this effect even more compared to strains and crosslinked strains.
The meat data should not per se be seen as post-peptide generation nitrosation proof. For N-terminal nitrosated (hydroxy)proline, it could be originally/partially nitroso(hydroxy)proline that is build into the amino acid chain. Considering the origin of gelatin theoretising based on meat data makes sense to me.

Considering the 1/3 chance that the N-terminus is (hydroxy)proline, the low potency of nitroso(hydroxy)proline, the fairly low ratio of MW (hydroxy)proline to MW gelatin, the typical low exposure to gelatin via medicines (definitely < 5-10 g/day (daily lifetime on average) I would say) and other considerations like the available testing data and reactivity considerations, I don’t think it is critical to know how meaningful in reality the reactivity of N-terminal (hydroxy)proline is in gelatin as worst case estimates on their own unlikely identify a risk (unless you think the meat data is an underestimation for some reason, e.g. 3D structure comparison and process comparison). The formal regulator general recognition of non-carcinogenicity of nitrosated proline (free or in a chain (that’s anticipated to get hydrolysed in vivo)) is still helpful to avoid having to distinct between the origin and faith of nitrosoprolines in risk assessments.

This is linked to side chains/the context of lysine, for N-terminal proline it is difficult to establish when the nitrosation if any occurs, in theory this doesn’t have to be after the original peptide structure was built.

As mentioned and explained before, I generally disrecommend total nitrosamines analyzer technology on peptides and this is supported by evaluation of ATNC data (published and unpublished) showing that often the risk for overestimating the possible presence for potent N-nitroso-compounds is too high, meaning that follow-up targeted testing was in many cases needed and complex to filter out both less potent nitroso compounds (like nitrosoprolines) and false positives. Only do this is you are well experienced with evaluating ATNC data on the equipment used (risk for detecting nitrite as well?) and are prepared to investigate significant positives for nitrosated compounds in general (while even in the targeted testing robustness limitations can apply as you will need to hydrolyse to test and need shown robustness on the individual amino acids first).

As we are mostly discussing nitrosamines here:
If somebody is aware of an example where the ATNC positivity on a peptide was well predictive to the later tested nitrosamine positivity (thus (hydroxy)proline related) on sample, I’m interested!
(Similarly: any proven negative ATNC’s on peptides?)


Thanks for your detailed answers into the topic. If we could define a common ground, what I could say is that there is enough evidence to dismiss the risk with a solid scientific but pragmatic background (e.g., regulators acceptance on -tive in-vivo results for proline, etc.).

For the sake of science I would love someone to do an study of nitrosation of gelatin capsules and its implications. But, pragmatically speaking I think resources (much more when are kind of strictly limited) should be more directed other well known nitrosamine root causes. To avoid drowning in a glass of water.

Or maybe, some months/years from now someone do the study and we find gelatin is a nitrosamine mess (full of meaningful mutagenic nitrosamines, etc). That is the kind of uncertainty the nitrosamine topics have behind. But until now, evidence (althought of course not abundant) I may say, indicates otherwise.

1 Like