In the previous blogpost the complexity of delivering inhaled drugs into the lung was discussed. It was shown that functional respiratory imaging is able to give an accurate measurement of the amount of drug reaching the lungs of patients with different respiratory diseases. The accuracy of functional respiratory imaging is not the result of some kind of witchcraft but can be explained by the guarantee that a maximum of real life measured parameters are used when evaluating deposition.
It is known that mathematical models are very capable of describing reality. At present any new aircraft or Formula1 car is completely designed, tested and optimized using a mathematical description of airflow (computational fluid dynamics or CFD). However, in healthcare, CFD calculations are very often corrupted by using estimations and guesses as inputs, often referred to as ‘garbage in, garbage out’. When developing FRI, the choice was made that only measurable values would be used as input when looking at lung deposition in patients.
The first, and by far the most important input when looking at lung deposition, is the patient. Each patient has his own respiratory tract geometry, and the variabilities found in these geometries are similar to the way you are a different human being as compared to anybody else. As can be seen from the image below, an average patient simply does not exist.
Figure 1: A series of lower airway models reconstructed from CT scans of 9 nonCF patients (in collaboration with Charles Haworth and Andres Floto at Papworth Hospital, Cambridge).
Knowing that a lot of variability exists between two patients, there are great risks if one would only look at an average airway geometry when trying to make statements of an entire population. These average airway trees that are often constructed, are a smoothened version of real lungs and lack the regional deformations typical for most of the respiratory diseases. Accounting for these variabilities in a population, is the reason why clinical trials are set-up as they are, selecting a representative sample from the population. Equally so, any method that wants to describe a population should include this variability. This is one of the main reasons why FRI uses real patients’ geometries that are extracted from HRCT scans.
However, not only the airway structure varies between patients. Another large inconsistency is the way they breath and the way this incoming air is distributing into the different lung regions. To give you an example: even when industry professionals were specifically asked to inhale the way they believe a Diskus DPI should be used, the differences are immense. The graph below shows the variability in inhalation manoeuvres of this group. Inhalation time, peak inhalation flow rate and the shape of the inhalation profile all show a large variability in the population.
Figure 2: A series of 75 collected inhalation profiles from people asked to inhale ‘as you think you should with a Diskus DPI’ at the stand of Coalesce at DDL a few years back. Breathing profiles were measured at the mouth (or device).
Not only the inhalation profile measured at the mouth (or device) is different, also the way the incoming air is distributing throughout the lung is very patient specific. For certain diseases, general trends are observed, but on an individual patient level the trends don’t necessarily hold true. This is due to patient specific differences in local airway resistance (inflammation, mucus plugging, …) or tissue compliance (emphysema, fibrosis, …). FRI uses this patient specific information on internal airflow distribution, combined with the relevant inhalation profiles to determine the deposition of inhaled products.
Secondly, also the device, and its characteristics play an important role. Each device is known to have its proper internal structure, resistance, particle release mechanism,… which all have an influence on the flow field in the oropharyngeal tract, and hence on particle deposition. The movie below shows how the internal structures and the resulting resistance of two different devices cause the flow to enter the patient in a completely different manner, even when the inhalation profile remains constant.
Movie 1: Velocity contours in an asthma patient breathing through 2 different devices (blue indicates no velocity, red indicates high velocity) using the same inhalation profile.
Another important input parameter that is related to the device is the spray plume. This plume is adding additional momentum to the airflow and carries the (solubilized) drug particles into patient. Both the aforementioned device characteristics (internal geometry, resistance…) and these plume characteristics are taken into account in the FRI deposition calculations to reflect reality as closely as possible.
By using all these inputs based on actual measurements, and by not relying on additional assumptions, FRI technology is able to very accurately calculate the internal flow that would arise in a specific patient when using a certain device. For this, FRI relies on computational fluid dynamics (CFD). CFD is a mathematical approach on solving the Navier-Stokes equations. These equations describe mathematically how flow behaves, based on the conservation of mass (continuity equation), momentum and energy. Nowadays CFD is the ‘go to’ technology in many different sectors such as aerospace, turbomachinery, combustion, oil piping, car industry,… The method is also making its way in the healthcare sector for non-invasive coronary artery disease detection.
Figure 3: Computational fluid dynamics: the Navier-Stokes equations (left) and some examples of outcomes in other industries (right).
Lastly, the drug itself needs to enter the virtually breathing patient. Particles will be released from the exact location in the device into the patient. These particles will experience forces linked to the flow inside the breathing patient which are in relation with their aerodynamic characteristics. FRI makes sure to inject particles with the right aerodynamic properties when calculating lung deposition. We use particle data that is obtained from aerodynamic particle size distribution measurements (ACI, NGI, laser diffraction, …)
FRI deposition is showing the same results as scintigraphy, but a mathematical model can only become a measurement when all the conditions (cfr. boundary conditions: the input conditions that the CFD calculations take into account when measuring deposition) are rightly chosen. FRI deposition consists out of 3 building blocks that guarantee the in-silico measurements to be as close as possible to the in-vivo situation. At FLUIDDA we continuously strive to improve our FRI tools in a mission to advance respiratory drug development and the treatment of millions of patients suffering from respiratory diseases.