AI for small nitrosamines not listed in EMA, FDA or Health Canada

Hello everyone,

I would like your help once again on the following dilemma we are facing as a team.
We are now dealing with small nitrosamines testing which are not listed in any of the guidelines but are included in CPDB (Carcinogenic Potency Database) and LCDB (Lhasa Carcinogenicity Database) and have a TD50 value.
Specifically I refer to the nitrosamines: 1-Nitrosopiperazine (NPZ), 1,4-Dinitrosopiperazine (DNPZ), N-Nitrosodiethanolamine (NDELA) and 1-Nitrosopyrrolidine (NPYR).
In some cases EMA or FDA used the TD50 value as AI. Is it safe to use TD50 value for these nitrosamines in order to extrapolate their limit or should we go with 18ng/day AI as a default?

For your reference the respective TD50 values are:
1-Nitrosopiperazine (NPZ): 8.78mg/kg/day
1,4-Dinitrosopiperazine (DNPZ): 3.6mg/kg/day
N-Nitrosodiethanolamine (NDELA): 3.17mg/kg/day
1-Nitrosopyrrolidine (NPYR): 0.679mg/kg/day

Thank you in advance


Hi, according to the EMA website, there is a method for calculating the AI from the TD50 value, taking into account the average bodyweight and the exposure timing.
To calculate a compound-specific acceptable intake based on rodent carcinogenicity potency data such as TD50 values (doses giving a 50% tumor incidence equivalent to a cancer risk probability level of 1:2). Linear extrapolation to a probability of 1 in 100,000 (i.e., the accepted lifetime risk level used) is achieved by dividing the TD50 by 50,000.

Acceptable Intake was calculated from TD50 as follows;

AI = (TD50 / 50,000) x 50 kg

The weight adjustment assumes an arbitrary adult human body weight for either sex of 50 kg. This relatively low weight provides an additional safety factor against the standard weights of 60 kg or 70 kg that are often used in this type of calculation. It is recognized that some adult patients weigh less than 50 kg; these patients are accommodated by the inherent conservatism (i.e., linear extrapolation of the most sensitive organ site) used to determine an AI.

Hope this helps.


Dear @manasipkane thank you for your feedback.

1 Like

Hi, @elenaly.

In addition to @manasipkane 's answer, the following @David 's post may help you. I think you can use not harmonic but each TD 50 for AI calculation.


One watch-out is to ensure that the cancer bioassay used to establish the TD50 is adequate. There is guidance in ICHM7 for what constitutes as “robust” bioassay, but the bioassay doesn’t necessarily have to meet all of the criteria described in that guidance to be able to serve as the basis for calculating an AI. That said, note that EMA established an AI of 26.5 for n-nitrosodibutylamine (CAS 924-16-3) based on read-across from NDEA even though there is a published TD50 for nitrosobutylamine that would support an AI of 691 ng/day. This is just a caution that not all TD50s are created equal and not all are sufficiently robust to serve as the basis for calculating an AI.


Thank you both for your feedback. @SusanFelter @Yosukemino

1 Like

Hi, In my opinion, you can use TD50 (use conservative value from CPDB / Lhasa data) to derive AI and apply control strategy. The main challenge is to prove that toxicophore is respective NI responsible for N-Nitrosamine reactivity. Thank you.


I think you can proceed with the T D50 value and get the limit by using the TD 50 values as now a days apart from listed nitroso impurities, authority is publishing some more impurities and those are not
covered in guideline.


Hi Mam,

Still question remains same, when we have to use the read across approach and when published TD 50 values, to decide the AI limit for Nitrosamine?

Could you please clarify me? Mam.


Unfortunately, it is not a simple answer. You can use the TD50 if it is from an adequate bioassay, but this takes a lot of judgment to determine. The ICH guidance provides a definition for a “robust cancer bioassay” but they also say that you don’t have to meet all of the criteria in order to be able to use a TD50. I am aware that the regulators have not always used published TD50s when they exist and it’s probably because they thought the bioassay was not sufficiently robust (e.g., dosing was for too short or there were too few animals per group or too few dose groups). There is no simple answer to this though and you will need to consult with a toxicologist who has experience with evaluating cancer bioassays to decide if you think the TD50 is sufficient to support calculation of an AI.


Read across and analogue selection is tricky work. So please always consult toxicology subject matter expert for AI derivation support. At later stage, He/She can justify analogue selection for follow up queries from regulators.