A recent Pharmacopeial Forum (PF ) article 1 aims to stimulate thinking on the methodologies used to calculate mass median aerodynamic diameter (MMAD). The article presents several MMAD calculation techniques as alternatives to the US Pharmacopeia (USP) method. Because our lab performs this calculation on a routine basis (and I publish a separate website dedicated to the calculation of MMAD ), I have a special interest in the topic.
USP <601> calls for plotting, on log probability paper, the percentages of mass less than the stated aerodynamic diameters versus the aerodynamic cut-off diameters. MMAD and geometric standard deviation (GSD) are determined from the plot. In practice, software (e.g. Excel) is often used to automate the determinations. The USP approach makes an underlying assumption that the mass-weighted data are log-normally distributed. The PF article argues that aerosols from many orally inhaled products do no meet this criterion, and thus the MMAD calculation should be performed using techniques that do not log-transform the data. 2
Based on a wide range of particle size distributions that I’ve observed in our lab, most deviations from log-normality are due to deviations at the edges of the aerodynamic particle size distribution. As pointed out in the article, the USP method gives equal weighting to these extremes, where there is little drug recovered and assay variability can be high. The practice in our lab has been to use the data points that represent a large proportion of the collected mass (e.g. those points closer to the center of the distribution), to calculate MMAD. One example of this approach, as prescribed by ISO 27427-2010 3
and used at MMADcalculator.com, is to use two points closest to and on either side of the 50 th percentile cumulative distribution, to calculate MMAD from the log-transformed data. Another approach we have employed is to use data from the two or three impactor stages having the largest amount of collected drug (which usually encompass the two points on either side of the 50 th percentile cumulative distribution).
I agree with the PF article’s conclusion that the USP methodology for MMAD calculation should be updated. And I do see the appeal of a simple, two-point interpolation of non-log-transformed data, which gives close agreement to the more mathematically cumbersome two-point log-transformed model. How do you calculate MMAD, and where do you stand on this topic?
All of this discussion should be kept in perspective: MMAD has less value (for example in the eyes of the regulators) compared to other parameters derived from cascade impactor data, such as the masses of drug on the stages and the fine particle dose.
References and Notes :
1) Christopher JD, Dey M, Lyapustina S, Mitchell JP, Tougas TP, Oort MV, Strickland H, Wyka, B. “Generalized Simplified Approaches for Mass Median Aerodynamic Determination”, Pharmacopeial Forum. Vol 36(3), May-June 2010. This article was published as a part of the USP’s “Stimuli to the Revision Process”.
2) The alternative calculations that do not use log-transformed data do not provide the GSD parameter. However, the PF article suggests that alternative metrics can be used.
3) Anaesthetic and Respiratory Equipment – Nebulizing Systems and Components”, International Standard ISO 27427, 2 nd Edition, 2010.