Issue link:

Contents of this Issue


Page 20 of 35

Inhalation April 2020 19 Looking further at the FPD data for these distributions (Panel C), we see that the current batch's FPD does not stand out at all from the pool of typical FPD values. In fact, the FPD for the current batch is just inside the lower quartile of the data set, meaning that it is within the bounds that define the central 50% of FPD values for this product. Although we do not know the precise FPD specification limits for this product, it is safe to assume that an FPD value between the upper and lower quartiles of this data set would conform to any realistic FPD specification. Otherwise, the manufacturer would be rejecting at least one out of every four batches (spe- cifically every batch whose FPD falls below the lower quartile, which comprise 25% of the tested batches), and very likely some batches above the upper quartile as well. Frankly, this would be an absurd and utterly unsustainable practice. In summary, the purpose of performing APSD testing on the current pMDI batch was to determine whether you had manufactured what you intended to manufac- ture, and, in the face of compelling evidence to the con- trary, FPD clearly "thinks" that you succeeded. It is hard to understand the value of APSD testing that concludes this batch is just like all of the others. Moreover, what is the point of training scientists, of qualifying chro- matography systems, of calibrating balances and flow meters, and of mensurating cascade impactor stages, if you ultimately rely on FPD to control your product? for each APSD are shown in Panel C with the same color-coding and these FPD values are also summarized with a box-and-whisker plot to show the mean and quartiles of the data set. It is worth noting that these APSDs are not simulated data; they are real QC test results for an actual marketed product (see sidebar for more detail). So let's return to the task at hand: you need to determine whether the current batch should be released to the European market. Essentially, you need to determine whether you manufactured what you intended to man- ufacture. At a glance, it appears that this should be a trivial decision. e APSD of the newly manufactured batch visually stands out from the other 200 APSDs. Moreover, examination of the cumulative plots (Panel B) of the APSDs reveals that the product's mass median aerodynamic diameter (MMAD) is typically between 1-2 µm, whereas the current batch has an MMAD of about 3.5 µm. Even in the absence of product specifi- cations (which were not provided by the pMDI manu- facturer), the current batch clearly does not perform as intended. Simply put, you intended to manufacture a product whose performance is summarized by the gray data, but you have failed to do so. It appears that disposition of the batch should be a straightforward matter: the batch does not perform as intended and should therefore be rejected. Unfortu- nately, if you were to rely on FPD alone to assess this product's performance, you would have reached the opposite conclusion. Figure 1 Panel A: The APSD of a single, newly-manufactured pMDI batch (red) is superimposed on 200 typical APSDs (gray) for comparison. Panel B: The same APSD data as Panel A, displayed as cumulative distributions. Panel C: FPD values for each APSD with accompanying box-and-whisker plot. 35 30 25 20 15 10 5 0 100 80 60 46 40 20 0 100 80 60 46 40 20 0 100% 80% 60% 40% 20% 0% %LC FPD (%LC) FPD (%LC) Cumulative % Mass Aerodynamic Diameter (microns) Aerodynamic Diameter (microns) Is this what you intended to manufacture? FPD thinks so! 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 A C B

Articles in this issue

view archives of Inhalation - INH0420