Tolerance interval
This approach may be used if adequate historical filtration data exists to calculate a tolerance interval and capture the
expected long-term behavior of the DS fill process. Typically, the tolerance interval contains 99% of the population (coverage)
with a 95% confidence limit. However, the confidence and population coverage depends on the size of the historical data set.
Care should be taken in combining data from laboratory or pilot scale filtrations with full–scale or manufacturing data, as
the non-recoverable volume of the systems may result in substantial differences in terms of batch uniformity. Additionally,
the calculated uniformity limits should not be wider than the DS/DP specification limits. While this method is based on actual
historical experience, it does require additional sampling and testing during clinical lots.
Equivalency acceptance criteria
The objective of the equivalency approach is to test the null hypothesis of non-equivalence (non-uniformity) within a DS batch.
If the null hypothesis is rejected, evidence of uniformity within a batch is demonstrated. Multiple samples are required for
each sample point within the batch, and the means are calculated for each sample location. Uniformity within a batch is demonstrated
when proscribed confidence intervals (typically at 90% or 95%) of the difference between the means are within the calculated
acceptance criteria. The benefit of this equivalence acceptance criteria (EAC) approach is a statistically defined proof of
uniformity within the allowable variability. However, this method does require a larger number of samples to be collected
from each sample point in order to have sufficient statistical power to make the acceptance criteria meaningful.
–EAC<µ1 – µ2<EAC
–EAC<µ1 – µ3<EAC
–EAC<µ2 – µ3<EAC
where:
µ1, µ2, µ3 are the mean of the sample test parameter over the course of bulk filtration (e.g. protein concentration at the
beginning, middle and end), and EAC is the equivalency acceptance criteria. It is beyond the scope of this paper to describe
equivalence testing; References 2 and 3 provide general sources on statistical equivalence.
|
