0:00
Yeah, thank you for the introduction.
0:02
Thank you for the organisers for inviting me today.
0:06
So yes, as mentioned, I'm going to be talking about in vitro dissolution absorption tools and using flux as a predictor of oral absorption.
0:16
So in this presentation, I'm going to give you an introduction to the in vitro tools that we have.
0:21
I'm going to provide you with some case studies.
0:25
Then we're going to look at predicting in vivo fraction absorbs, in this case human fraction absorbs based on the in vitro flux profiles.
0:34
We'll look at the evolution of those models, where they worked and some improvements to the models.
0:40
And I'll just finish off with some conclusions.
0:45
So Pion was actually the first company to commercialise the PAMPA technique, the parallel artificial membrane permeation assay, high throughput method for rank ordering during the permeability of compounds.
0:58
It was invented by scientists at Roche and Pion worked with those scientists and was the first company to commercialise the platform.
1:05
It looks at the permeation across a membrane barrier.
1:09
So we have a phospholipid plated onto a filter.
1:13
You put dissolved drugs into a donor plate and after a period of time you measure how much drug has appeared in the receiver and you can calculate the permeability.
1:21
So it's used in drug discovery during lead optimization for rank ordering and selecting candidate compounds for further development.
1:30
And Pion optimised the Kansy method and ended up with a pretty good correlation to human jejunal permeability.
1:41
And we still supply this method now.
1:45
However, the overall drug absorption process is not just looking at permeability, it's a composite of the solubilisation as well as the permeability to give you the overall absorption process.
1:59
So drugs need to be in solution before they can be absorbed.
2:03
They then have to cross the membrane barriers.
2:05
This means they need good physical chemical properties needing good solubility and also good permeability to provide you with a good total oral absorption.
2:18
And so we developed various tools which we can use throughout the drug development process from low volume compound sparing assays where we've made some donor compartments separated from an acceptor compartment by the same phospholipid type membrane that was representative of human genuine permeability.
2:40
Can be used in the preclinical preformulation early development setting through to larger scale assays when you go in further into formulation development through to integrating absorption chambers into larger United States pharmacopoeia type vessels for final formulation development, analysing, full dosage forms such as tablets and capsules and even using these techniques to do some bio in vitro bioequivalence determinations by comparing products.
3:10
So we can integrate an absorption chamber where we have a membrane on the bottom, you can put your full dosage form in, and then you can measure the combined dissolution absorption performance.
3:22
And the way we typically do this is to do direct in situ concentration analysis using our rainbow spectrometer.
3:31
And that's this platform here.
3:32
It's a fibre optic spectrometer, parallel probes up to 8 fibre optic channels.
3:38
We have a deuterium lamp.
3:39
We shine UV light along with fibre optics.
3:42
There's a open window at the end, so you can put this into your dissolution medium or into your absorption chamber.
3:49
Light reflects off a quartz mirror back along return fibres to some Spectra photometers and so we can do direct in situ concentration measurements and so we can build up an information rich detailed profile about our dissolution and absorption process.
4:04
In fact, we can capture Spectra every two seconds, every 5 seconds if needed.
4:13
And just a little bit more about the membrane.
4:17
So we optimised A membrane or you know an artificial membrane by using a mixture of phospholipid components, some neutral, some negatively charged.
4:30
We supplied this in ampules which you can store in the freezer and then we coat them onto the filter support in our absorption chambers.
4:37
And you can tell this creates an integral membrane barrier because the whole thing goes translucent.
4:45
And so we have a intact membrane barrier which we can then put these absorption chambers as I say into dissolution vessels.
4:53
We typically on the receiver side have something they call except a sink buffer to maintain sink conditions.
5:03
So that's a bit about the kind of apparatus that's look a little bit more into flux definitions and some case studies.
5:14
So the definition of flux, flux is a measure of transport.
5:18
It's the net number of molecules or massive particles crossing a unit area of membrane per unit time perpendicular to that unit area.
5:28
And so it's described here, it's the mass of molecules per unit time divided by the area is the flux.
5:37
But that can also, I mean, of course, it's difficult to measure, it's virtually impossible to directly measure inside the membrane, the mass transport.
5:46
But what you can do is you could rewrite that expression by measuring what appears in the receiver compartment, so how much mass actually transports through, and you can represent it as the concentration, the appearance of the concentration over time multiplied by the volume in the acceptor chamber divided by the surface area of the membrane.
6:07
And so the flux is then proportional to the slope of the appearance concentration in the receiver compartment.
6:14
And units are typically in mass per unit of time per square centimetre area of membrane.
6:22
And of course, this can also be rewritten according to using fixed first law as being the effective permeability through the membrane multiplied by the donor concentration.
6:34
So the concentration in the donor compartment, and that's just fixed first law.
6:42
So let's look at some examples case studies.
6:47
So here we have two preparations of carbamazepine, a poorly soluble drug.
6:52
We're using the microflux apparatus.
6:54
So we have donor compartment separated from acceptor compartment by the biomimetic membrane and we have poorly soluble carbamazepine.
7:03
We have the API alone.
7:04
This just shows the donor compartments.
7:06
I'll show the acceptor compartments on the next slide.
7:10
And so we have one milligramme per millilitre load in each case.
7:14
We have a API alone and then we have a Soluplus preparation.
7:19
So for the API alone, we only get about 1/4 of the dose dissolves and then there's a slight precipitation event.
7:25
Carbamazepine is known to form several different polymorphs, so it probably precipitates down to a less soluble polymorph and then it maintains the plateau solubility with the Soluplus preparation.
7:37
Here we have an example of the kind of the spring and the parachute effect.
7:40
So the polymer is able to fully dissolve the whole days.
7:44
You know it gets up to the one milligramme per millilitre concentration and then there's the parachute.
7:50
So the polymer is acting as a precipitation inhibitor and preventing precipitation.
7:55
And if you look at the time scales, you know this parachute effects is persist for, you know really quite a long time.
8:02
So how does that translate to flux profiles?
8:05
So here we have the concentrations in the receiver compartment, in the acceptor compartments and we can plot the concentration versus time.
8:14
And then the flux is just proportional to the slope of the appearance line.
8:18
So if you multiply that by the volume of the acceptor compartment and divide it by the surface area of the membrane, you actually get the flux values.
8:26
And so the API alone, we get a flux of about 1 microgram per minute per centimetre squared.
8:32
And for the solid plus preparation, the flux is actually about 2 1/2 times higher.
8:37
So the higher concentration in this particular case is providing a driving force for absorption across the membrane and we get a much higher flux.
8:48
OK, that's all very well.
8:50
So, why do we need to measure flux?
8:53
I can see my concentrations higher in the donor compartment.
8:56
Why don't I just do dissolution?
8:58
Do I need to do flux measurements?
9:01
Well, there are many different examples depending on the types of formulation ingredients that you might use where you can actually lead to a suppression in the flux.
9:09
And we have here an example of a surfactant formulation.
9:12
So we have the raw unformulated API which has this dissolution performance in the donor compartment, and all this data is measured using the fibre optic probe.
9:22
And then we have a surfactant suspension which dissolves almost immediately to, maintains or fully dissolves the dose.
9:31
But then when we look at the flux profiles and we see what appears in the receiver compartment, it's actually the unformulated API that has the best flux performance.
9:40
We actually have suppressed appearance of the compound from the surfactant suspension.
9:48
So this is an example of flux suppression.
9:50
And this, this can depend, on individual APIs and different types of formulation ingredients.
9:58
And what's known to happen here or what's suspected to happen here is micellar entrapment.
10:04
The surfactant micelles, the, the compound is bound or trapped from inside those micelles.
10:09
Those micelles are too large to move through the membrane.
10:13
It's only free monomeric drug that can be dissolved and can permeate across the membrane.
10:20
And so we actually end up with a flux suppression effect.
10:23
And that's also been demonstrated in the literature.
10:32
So here's some examples of etoposide measured by a group at Ben Gurion University in Israel.
10:39
They looked at the dissolution absorption performance of a range of different excipients from a cyclodextrin to a surfactant to K solvent and an amorphous solid dispersion.
10:52
And most of the examples while giving solubility enhancement lead to an apparent decrease in the permeability.
11:00
They present it as an apparent decrease in the permeability.
11:04
And again, so this was evidence of the solubility permeability interplay coming into effect.
11:10
It was actually the amorphous solid dispersion showed the best performance here.
11:16
It is free monomeric dissolved drug.
11:19
It's you know, it's the amorphous solubility, it's providing the increase in the concentration and the driving force for the absorption process.
11:27
So we believe it's important to do these types of analysis to understand the composite effect of solubility and permeation on your drugs performance.
11:39
But we wanted to take these types of new studies further and see what else can we do with the data.
11:46
And so we've developed the software module called Pion Predictor, where we can take the flux data and use it to try to predict fraction of drug absorbed.
12:03
So we did this by working with Kiyohiko Sugano used to be at Pfizer, but he's now back in Japan at Ritsumeiken University.
12:14
And he came up with something called the GUT framework, which stands for Gastrointestinal Unified Theoretical Framework.
12:22
And so we've now tried to implement this using our flux data to predict fraction of drug absorbed or oral fraction of drug absorbed.
12:31
And I'll try to go through this in the next slides.
12:35
So we have our in vitro data and you know, our flux appearance profiles from our experiments.
12:43
We import this into the software.
12:45
We can calculate the effective permeability from the donor compartment concentration.
12:50
And then we set up other relevant information such as the species model.
12:54
You know, ideally we do this for human, but we can also introduce parameters for other species such as rats.
13:01
Then we model this data, fit a profile and then we can use it to try to predict the fraction of drug absorbed.
13:09
And I'll try to explain that in the next few slides.
13:15
So mass absorbed from the in vitro assay is essentially, for a given intestinal transit time and surface area, the amount of material absorbed across a membrane can be determined by multiplying the flux by the surface area of the membrane multiplied by the transit time.
13:32
You know, the total mass absorbed would be the flux times surface area membrane times transit time.
13:38
But what we need to do is to be able to scale this up to the particular species from an in vitro setting, for example, to human.
13:46
You know, we know human has a much larger surface area of membrane, for example, than our in vitro test assays.
13:53
And then the fraction absorbed is just simply the, the total mass absorbed divided by the dose multiplied by 100.
14:02
So in the GUT framework, we take our in vitro flux data, and we need to scale it up to the in vivo relevant flux.
14:13
And we need to do this by determining the total surface area available for absorption for the particular species that we're modelling, for example, human.
14:23
And so we convert out in vitro flux profiles, we scale them and then based on the observed permeability in the source assay, we then scale the flux relative to the degree to which an API can permeate through the intestinal membranes.
14:43
Now the amount that can do permeate through intestinal membranes actually does depend on the permeability of the drug.
14:50
So it's important to know that parameter and that's because you the intestinal membranes, if you look at a picture inside the intestine, it consists of a series of circular folds and then villi structure, it's the villi structure that give a huge surface area for the intestinal membrane.
15:11
But depending on the permeability of the molecule they can access, they can have different levels of access to the different parts of the membrane.
15:20
And compounds with high permeability, these tend to only, these are absorbed, pretty much through the, the circular folds from the top parts of the membrane.
15:31
And it's only compounds with low permeability that are able to diffuse through entirely to the villi structure.
15:38
So you end up with different surface accessibility for compounds depending on their permeability.
15:44
And so these are the types of things that need to be incorporated into the model.
15:52
And then we ended up having to apply a scaling factor for compounds with intermediate permeability.
15:59
And so this was also done.
16:01
And we'll see, we'll see some of the effects on the models in, in the next slides when I go through some of the examples of how well the models performed.
16:14
So let's start to look at some examples of how well using in vitro flux data predicted human fraction absorbed.
16:23
So here's the next slide is the first evolution of the model.
16:28
And actually it didn't do too well.
16:32
So we didn't use the scaling.
16:36
We used a, a kind of broad classification of high permeability compounds with only have access to the circular folds and low permeability compounds would have access to the entire membrane.
16:48
And so we had a complete cut off.
16:51
This didn't work particularly well.
16:54
You can see here the graph shows, you know for these sort marketed products we have the calculated fraction absorbed versus the human in vivo fraction absorbs the unity lines drawn in blue.
17:08
The correlation line of our data versus a prediction is the orange line and you know an R square shown down there.
17:15
So it didn't do particularly well, so the broad cut off of the classification between permeabilities didn't work so well.
17:25
And also over prediction is quite an issue.
17:29
You know, we're predicting actually 500% absorbed, almost 460% absorbed for one of these drugs, which obviously doesn't make sense.
17:39
So in the next evolution of the model, we applied the scaling and also we made some corrections to the difference in what's perceived to be the unserved water layer, which is a concept when you come close to a membrane barrier, there is a stagnant water layer which can also act as a barrier to permeability.
18:01
So we made some corrections for what we perceived to be in vitro unserved water layer versus the in vivo situation.
18:10
Well, it has improved things.
18:12
You know, we've got a better R-squared.
18:14
You know, we have 3 compounds now instead of just two that are close to the unity line shown in blue.
18:21
Intuitively, some of the things they're doing, are doing it right in the, we have several formulations where we have a low dose and a high dose and, the lower dose is, having a higher fraction absorbed, which, which could make sense for a compound with poor solubility or poor flux.
18:40
So there's a couple of situations here with a low dose and a high dose.
18:44
I think here at a low dose and a high dose, which again, showing the lower dose has a, has a higher value than the high dose.
18:54
But again, it's not brilliant, but it's improved the model and over prediction is still an issue.
18:59
We're still going, in this case almost up to 200% as it are calculated fraction absorbed. Well in the third model.
19:09
What we needed to do was make a correction for what's not happening in the in vitro assay is the amount of drug removed is not being totally reflected in the in vitro assay compared to what would be in vivo's situation, which once the drug has permeated and then it's removed into drug circulation, whereas that's a much longer kind of process in our in vitro assay.
19:35
So what we then did is made a calculation of the relative proportion of the dose remaining in the intestinal fluid or in the donor fluid at each time point, allowing for a proportional reduction of the calculated absorption rate, situations where intestinal removal would be high.
19:51
And now we're getting better.
19:53
You know, R-squared's improved.
19:56
We now have five of the compounds close to the unity line.
20:01
And the, in the orange line shows the correlation, but there's still a couple of cases where we're not doing so well.
20:10
And so we, we kind of went back to the drawing board and thought about, what might be happening here.
20:17
And there's a concept known as the particle drifting effect, which was also published by Kiyohiko Sugano, and published back in 2010, which basically says for small particles or high doses and nano suspensions and high doses, it's possible for particles to enter into the unstirred or stagnant water layer, which then provides some kind of reservoir for diffusion and release of drugs directly to the mucus, which could aid the overall absorption process.
20:51
And so this was referred to as the particle drifting effect. And the celecoxib, if I go back a slide, the celecoxib was kind of under predicting and these formulations of celecoxib, the marketed formulations are known to contain a significant amount of significant proportion of nanoparticles in the formulation.
21:16
And so we incorporated an improvement to the model to account for the particle drifting effect.
21:22
And, and we did this by, we can measure this in vitro by doing a series of different flux experiments at different dose levels to try to understand in vitro this particle drifting effect.
21:32
And then we can implement them into the model.
21:35
And actually now we have a pretty good correlation.
21:38
You know, all molecules are within ±15 percent accuracy following the unity line.
21:45
The R square is extremely good.
21:47
OK.
21:48
It's a limited number of formulations.
21:50
It's a limited number of molecules, but this is the kind of version that we have in the software that we should be releasing in the next month or two.
22:00
We're still encouraging people that may have access to our flux tools already to come and discuss with us if they want to be a beta test site.
22:08
We obviously need to get more examples and more data, but I think what we've shown so far is, the results can demonstrate the applicability of in vitro flux measurements for the prediction of fraction absorbed for orally administered drugs.
22:24
We have these different flux tools, low volume, through to the late stage, large volumes where you can integrate an absorption chamber into USP dissolution apparatus.
22:34
So they're available all stages of drug development to guide formulation design and performance.
22:40
And as I mentioned, we need more data now over a larger sample set of APIs and formulations to test the robustness and further test the viability of the model.
22:52
So where next? OK, yeah, more data obviously.
22:57
But you know, we could start to combine the data with in vivo clearance elimination, and other PK parameters to simulate plasma concentration time profiles, it says drug concentration versus time.
23:10
Or we could start outputting resorption rate constants, for example, from our software to provide an input into other third party PBPK modelling tools such as Gastroplast and the SIM SIP, for example.
23:26
And with that, I'd like to thank everybody for their attention and I'm happy to take any questions if we have any time left.
