This is done by compiling data generated in different parts of the drug development workflow, and first analyse it with mathematical models, then connect these models into an integrated overview where the models describe the mechanisms of the system we want to understand. These models can then help guide new experiments and tests, and form the basis of future applications to agencies controlling the approval and quality of pharmaceuticals. This could save both time and money, and get new drugs out to the patients faster and cheaper. In addition, it could reduce the need for animal experiments. In this project, we will test and develop this new workflow through the testing of a specific drug, dapagliflozin, which is used to treat diabetes and to reduce the risk of cardiovascular complications, such as heart-attacks of congestive heart failure.
Type 2 diabetes (T2D) is one of the most common and expensive diseases in our society, and also one of the most important risk factors for the most frequent cause of death; cardiac disease. Although this connection has been known for a long time, it is not until recently that a new drug for T2D has been found – empagliflozin – which can decrease the risk of cardiac disease. In addition to this, it has been discovered that diabetes should actually be divided into 5 different subgroups.
AstraZeneca, the biggest pharmaceutical company in Sweden, has recently released the new and similar drug, dapagliflozin, but have not yet investigated whether this drug also decreases the risk of cardiac disease. If dapagliflozin has these positive side effects as well, it could widen the field of application of the drug, which would benefit both the company and diabetic patients.
To investigate this question using the traditional way of testing drugs would cost billions. It is therefore both important and exciting that FDA, the institution that certifies drugs in the USA, just opened up for applications for suggestions of a new type of certifications: certifications that use mechanistic, mathematical models.
The concept of mathematical models is not new, they are used as pharmacological models in every pharmaceutical company today, to predict how much of the drug is present in the blood. But Gunnar Cedersund’s special expertise – mechanistic, or systems pharmacological, models – describing both the intracellular mechanisms and the whole body level at the same time, are so far very uncommon. This is why Gunnar has been financed by AstraZeneca to use his systems pharmacological models to solve specific, but smaller, problems. One of these solutions led to the project winning AstraZeneca’s big internal research award IMED. Cedersund’s models have also won several other awards, both in clinical and systems pharmacological contexts. Despite this, neither Cedersund nor anyone else have yet developed a systems pharmacological model comprehensive enough to convincingly answer the question of whether or not dapagliflozin is efficient for preventing cardiac disease.
In this project, Cedersund will work with AstraZeneca in order to develop such a model, which additionally would be able to differentiate between the 5 subtypes of diabetes. Such a model would not only be of service in this specific case, but could be working as a knowledge base also for much of the future drug development that will take place in this area. Integrating all available knowledge and data in an interactive simulation model will serve many different purposes. In a pharmaceutical project one might for example ask: “Given all the available knowledge and data, what is the most probable scenario?” or, reversed, “What would be implausible, given the data we have today?” Such a knowledge database would also transform the linear drug development workflow we use today into an integrated whole, where different projects can help each other. Instead of a failed Phase 2 trial being waste money, it turns into valuable knowledge in other projects using the same knowledge database. In this project, we will build such a knowledge database by connecting a number of existing models, and use unique and extensive data from several research groups worldwide. To answer the above question about dapagliflozin, we will test two different hypotheses:
- H1: Both empagliflozin and dapagliflozin inhibit reflux of glucose from the urine, and thereby lowers the body’s sugar and energy levels. This also leads to a lower blood volume, which in turn lowers blood pressure, and decreases the risk of congestive heart failure. The first hypothesis is that these mechanisms on their own can explain the measured effects, because less energy means that the fat cells can store all the energy as fat, which is good because fat storage in other tissue leads to inflammatory responses and the complications present in congestive heart failure. Increased fat metabolism in the liver also leads to the release of ketone bodies, in turn leading to increased heart function.
- H2: The second hypothesis is that i) empagliflozin also has direct effects in other cell types, but they are the same as dapagliflozin, ii) with these effects added to the model, it can explain why empogliflozin has positive effects on heart diseases. We will measure such direct effects from both empagliflozin and dapagliflozin, both by adding them to isolated cells from real humans, and adding them to artificial human organ systems.
If H1 or H2 are true, and if the model can correctly predict key results that can be tested in new experiments, the model could in principle be the basis for a new certification application to use the diabetes drug dapagliflozin also for cardiac diseases. In that case, we will also establish a collaboration with the Food and Drug Administration in the US, and investigate how such a model could be used according to their new guidelines for model based drug certification processes.