Performance Characterization of EDA and Its Potential to Improve Decision Making in Product Batch Release - Good Cascade Impactor Practices, AIM and EDA for Orally Inhaled Products - page 200

Fig. 8.23 Identification of target cumulative APSD for CFC suspension MDI w9j601 : fit of logis-
tic model to cumulative mass-weighted APSD; %LC % label claim; “APS” in equation = aerody-
namic particle size (μm)

MMAD value, and its pseudo R 2 values were computed as a function of the displaced
MMAD . Limits in terms of MMAD were then established by selecting arbitrary R 2
values (0.9, 0.8, 0.7, and 0.6) and selecting the corresponding MMAD values as the
assumed limits. The calculation for the pseudo R 2 is given by Eq. (8.5) for clarity.

n n

å å

2 2

( yy - ) (% LC - % LC )

i i Displaced Reference

2 i = 1 i = 1

R = =

(8.5)

n n

å å

( yy - )

2

(% LC - % LC )

2

Displaced

i i Displaced

i = 1 i = 1

It is important to note that the differences were calculated for particular fixed
particle sizes as measures of distance between the curves vertically. This is one
approach for determining a bound around the target/reference APSD curve that
defines a range of acceptable MMAD values.
Again the reader is reminded that these selections are not connected in any man-
ner to the actual quality limits for the studied products. They were selected exclu-
sively to study the characteristic of the APSD-related metrics. Figure 8.24 illustrates
this process for OIP w9j601 when R 2 is set at 0.9. The entire set of limits (Table 8.6 )
employed for constructing OCCs were obtained via this process.
Using these assumed limits for the MMAD values, both cumulative APSDs
(Fig. 8.25 ) and APSDs plotted in the form of individual values of API per stage
(Fig. 8.26 ) could be constructed for hypothetical APSDs at the assumed limits. Note
that this exercise considers the case of a shift in MMAD of the distribution without
a change in its shape.