Product Life Cycle Approach to Cascade Impaction Measurements - Good Cascade Impactor Practices, AIM and EDA for Orally Inhaled Products - page 137

(f) Establish suitable precision for both LPM/SPM and ISM determinations
(which is also necessary for full-resolution CI). Many replicate measure-
ments by both full-resolution CI and AIM-QC system, ideally from several
different batches or at a minimum from widely spaced intervals during man-
ufacture of a single batch, will be needed in order to assess how small of a
change in MMAD and total mass entering the sizing part of the impactor can
be detected by the EDA metrics for that particular product.
(g) Quantify the minimum number of batches required to achieve a statistically
robust correlation between EDA metrics obtained from the full-resolution
and the chosen AIM systems. In line with the QbD philosophy, the OIP
sponsor will likely need data from many different batches (suffi cient to rep-
resent adequately sources of variability in the manufacturing process, input
materials, analysis and stability effects), with multiple CI runs per batch
(suffi cient to assess within-batch variability) in order to get a good estimate
of the target distribution (mean and variability), which would be representa-
tive of future production batches at release and for stability testing. The fol-
lowing outline provides an idea of what may be required:
(i) Establish the APSD of the product, which will require a large body of
full-resolution CI data.
(ii) In both the product release and stability programs run abbreviated
impactor based measurements in parallel to the full-resolution systems
to establish correlation between EDA metrics from both abbreviated and
full-resolution systems to enable use of AIM for routine control later on.
In totality, the data required in part (ii) will likely amount to hundred(s)
of CI measurements spread over numerous batches, refl ecting different
lots of API, device components, etc. However, the cost of this upfront
work should be more than offset later by the option to use EDA in con-
junction with abbreviated CI measurements in the QC environment.
This sequence of events mirrors the principle underlying the Quality-by-
Design approach, in which the full resolution CI essentially maps out the
“design space” for the product APSD, with the abbreviated CI working
within the “control space.” Such a regimen will also improve decision
making by virtue of enabling more samples from the batch under con-
sideration to be assessed for a given expenditure in terms of effort and
equipment.
However, despite the advantage of the approach just outlined, it is rec-
ognized that some companies may chose to collect these data only after
the product has been approved. In such instances, the switch from full-
resolution CI measurements with traditional data assessment to AIM-
based CI determinations coupled with EDA could still take place.
Nevertheless, it should be recognized that delaying this decision could be
associated with some business risk, because complete understanding of
the APSD-properties of the product in both measurement regimens (with
suffi ciently different batches) is also postponed.