The Application of Multivariate Data Analysis to Inform the Development of New pMDI Valves





With continually evolving orally inhaled therapies for the treatment of asthma and chronic obstructive pulmonary disease (COPD) and, more recently, diabetes[1], rapid development of highly effective new pressurised metered dose inhaler (pMDI) delivery devices is exigent. Further, with environmental concerns surrounding the global warming potential (GWP) of current propellants such as hydrofluoroalkane (HFA) 134a and 227[2], an additional challenge lies with maintaining the performance of pMDI inhalers with the introduction of new propellants. The present work describes the application of a statistical, multivariate data analysis technique to identify the exact relationships between aspects of inhaler design, formulation properties and the resulting dose delivery performance (as monitored by the shot weight) of the system.  Partial Least Squares Regression (PLSR) was first used to identify the relationships between the dependent and independent variables and from this, linear, beta coefficient equations to quantitatively relate the componentry and formulation properties to dose delivery performance were extracted. This led to the generation of regression prediction models to allow the specification of a pMDI valve based on its required performance and the properties of the formulation employed. Fidelity of the prediction was assessed by comparing predicted values with historical experimental data. Formulation constituents were found to have a profound effect on shot weight, with ethanol fraction being a very significant contributor. Whilst the exact mechanisms underpinning this are the subject of further investigation, the relatively low density of ethanol compared to HFA 134a is known to be influential and the effect of ethanol content on the vapour pressure is also thought to be a significant factor. 


Key Message


PLSR revealed the most important design and formulation factors affecting shot weight performance of Consort Medical pMDI valves and led to the development of a predictive model for new valves.