Theoretical 'Purge Factor'

In a recent FDA Q&A public session the following was discussed and debated: What is the scientific rationale behind the statement “Theoretical purge factor calculations may overestimate purging factor of the process.”

Theoretical purge calculations are designed to be conservative. However, the purge factors (i.e., 1, 10, 100, 1000) should be based on scientific evidence. In some cases, firms assumed that NDMA and DEA were “completely miscible” in water and therefore would be adequately removed from the process due to the high number of aqueous work ups and yet the API was contaminated with nitrosamines. Solubility studies discovered that while NDMA and NDEA were indeed “water miscible” they were also very highly soluble in organic solvents. The aqueous wash steps were much less effective than predicted at removing the nitrosamine impurities.

1 Like

A nice scientific paper addressing what purge factor could be used for such manufacturing steps of washing, phase separations, etc. based on the partition coefficents is the following one: Prediction of N-Nitrosamine Partition Coefficients for Derisking Drug Substance Manufacturing Processes


@Diego_HM that’s a great reference paper … I believe the applicability of the publication goes beyond risk assessment. Specifically on sample preparation for Nitrosamine Testing.
Can you share your perspective as to where we are in LATAM related to Nitrosamine? We saw yesterday’s publication on ANVISA’s guideline, but who’s next?

Hi Naiffer, regarding your question about LATAM. From what I have seen, Regulatory Agencies in LATAM have prepared working groups for specific Nitrosamines issues like Metformin (e.g., Chile), Ranitidine (e.g., mandatory or preventive recalls in Colombia, Chile, etc. for Ranitidine), but nothing more. In general, the most evolved regulatory frameworks under my perspective comes from Chile, Brazil and Mexico. However, I do not expect anything like a guideline for any other Health Authority in the short or even medium term, as this is a very complex topic that is constantly evolving, but who knows.

1 Like

@Naiffer_Host @Diego_HM
I want to confirm it. LogPow of NDMA and NDEA are -0.50 and 0.48, respectively. When both are extracted with water and 1-octanol, roughly divided into 3:1 and 1:3. Those will be simply assigned as purge score 3 and 1 from dilution. I’m surprised with it because NDEA is not purged well, despite water miscible nature. Of course we should consider lots of other factors for calculation. Anyway NDMA and NDEA are soluble in both water and organic solvents.

I want to add further comments to purge factor. In order to justify option4 strategies of nitrosamines with purge factor, high purge score is required due to extremely low acceptable intake. Besides, more than 100 of PR ratio is needed due to conservatism. When we consider nitrosamines purge in API manufacturing processes, we will find nitrosamines are less reactive and difficult to volatile. Therefore we expect purge by solubility, however purge by solubility should keep conservatism. I add the slides of @AndyTeasdale. This means multiplication of solubility purge in the same work-up processes are not preferable without the evidence of purge.

Of course the amount of nitrosamine formation and the number of manufacturing processes or extraction unit processes after formation is important. However in many cases, it is challenging to achieve enough purge score only by solubility without some evidence, I think. In addition that, nitrosamines with some alkyl chain or aromatic ring get lost the solubility in water easily.

On the other hand, nitrosating agents and nitrosable amines are easily removed, compared to nitrosamines. As @MichaelBurns demonstrated on his paper, purge score of both tends to get high even in small numbers of processes.That is why I recommend to justify less reactivity for nitrosamines formation due to some conditions such as amounts of both residual reagents, pH, temperature and pKa of nitrosable amines.

What do you think about the issue?


When it comes to applying solubility-based purge arguments it is important that users are very careful with their application, as the sheer volume of operations that may utilise solubility for the purpose of purification can cause the rapid escalation of that term - as @Yosukemino points out. At present we are drafting a manuscript with a number of the Mirabilis consortium members to define a best practice to this end. One of the principles we are putting forward is that evidence to support a purge assessment should relate to the importance of the terms used across the total calculation. If solubility is the most important factor within a calculation, then there becomes a necessity to back that assertion with some form of experimental evidence. The type/level of evidence required would itself be determined by the purge ratio.

1 Like

Thank you for your insight. I appreciate your help. In purge calculation, there are lots of patterns in using solubility score. And I agree definition of term is helpful to avoid confusion, it is sometimes complex and misleading. Of course I should pay attention to the number of purge ratio. I’m looking forward to seeing your new paper in future.

Also, in terms of understanding potential reactivity of nitrosamines, the review I have been involved with as part of the IQ consortium has now been published.


The excellent paper, Establishing Best Practice for the Application and Support of Solubility Purge Factors(, is published. @MichaelBurns and others described the rules of the purge factor related to solubility. It is helpful to avoid miscalculations at risk assessment of nitrosamine impurities. And we can change the supporting evidence depending on the purge ratio and dependence on the solubility score.


The latest version of mirabilis enables us to calculate the reactive purge factor of nitrosamines and secondary amines in manufacturing processes automatically through a new approach, the condition approach.

Development of the Methodology for in Silico Reactivity-Based Purge Predictions: Making Mirabilis Think Like a Chemist (


It is more consistent with how a chemist would approach an assessment.