I was wondering if a “Limit Test” strategy could theoretically be used to conduct NDSRI risk assessment in drug product to demonstrate that batches do not show any identified potential NDSRIs at or above 10% of the AI limit set forth by the regulatory agency? This of course would be provided that one demonstrates specificity and an appropriate lower limit of detection as per ICH Q2 guidelines.
Essentially, I am thinking one could perform sample preparation on the drug product “as is” and a replicate sample that has been spiked with a standard of the NDSRI at the 10% of AI level. Then analyze the samples with the analytical method and determine whether the drug product “as is” has a peak area at or above that of the spiked sample. I’d expect this should also help account for recovery and matrix effects that may be encountered?
I feel as though this would be scientifically sound and would avoid some of the additional qualification needed for the analytical method, but wanted to see what others think…
Yes, it is possible to use it, and it is a very good approximation. It is very useful when the nitrosamine comes from an impurity from the API and the risk assessment has identified a low/medium risk, and the analyses are performed to rule out the risk of NDSRI formation with greater certainty. I do not recommend using the limit test when the NDSRI comes from the API itself, as there is a high probability that the 10% limit will be exceeded and it will be necessary to quantify the amount of NDSRI.
To ensure that the sample is below the 10% limit, the response of the analyte in the sample must be less than half the response of the analyte in the spiked sample at the reporting threshold.
Hi, yes, it is possible. It ultimately depends on the level of risk you are willing to accept, as previously discussed. If you assess that the risk of the nitrosamine exceeding 10% of the acceptable limit (AL) is low (e.g., due to a low MDD, a Category 5 compound, and the relevant secondary amine being only an API impurity) yet you still want experimental confirmation, then performing a limit test is a reasonable approach.
However, be cautious when the secondary amine could also arise as a degradation product (i.e., controlled in the drug product) and not solely as a process impurity (controlled in the API). In such cases, you need to be prepared to answer questions regarding how close the detected signal would be to the 10% threshold if it were present. Sometimes it is useful to spike at two levels, such as 5% and 10% of the AL. If all results remain below 5%, you would typically not need to justify analytical variability or potential increases over shelf‑life, as long as representative batches covering the full shelf-life have been tested.
That said, achieving the required sensitivity is often the most challenging part of method development. Finally, remember that ICH Q2 requirements represent the minimum expectations. If, for your limit test, you consider that an additional validation parameter (e.g., repeatability) is necessary, you are free to include it. We have seen cases where going slightly beyond the minimum helped avoid issues later during routine testing.
Thanks! That makes total sense. In the case I was thinking of applying this, we are looking at some tertiary amine APIs with a possibility of formation of the desmethyl NDSRI. We believe the chances of us seeing anything at or above 10% of the AL are low, so a limit test seemed like a good first step to assess where the DP is with respect to that threshold. We figured that if we do get a result where it’s too close to call or clearly over the limit, we can always go through the process of doing a standard curve and evaluating where exactly the level is.