QbD Implementation on Medical Devices (Drug Eluting Stents)
Does Quality by Design apply to Medical Devices?
We’ve seen it coming down to combination products. Now QbD implementation on Medical Devices. My colleagues working in the medical devices claims Design Controls is comparable to QbD. Design controls is a FDA guidance to product development activities. However, it is an outdated approach resembling the waterfall or stage-gate product development approaches. Both QbD and Design Controls could certainly benefit from each other.
QbD Implementation on Drug Eluting Stents
Here is a sign that QbD may be trickling down to Medical Devices. Dr. Ken McDermott and Dr. Sharmista Chatterjee of FDA present “Application of the QbD (Quality by Design) Approach for Coating Drug Eluting Stents (DES).
If you are not familiar with Drug Eluting Stents, you will be surprised how common it is for my friends and families to have one or more in their bodies. In fact, 1 million stents are used every year in the US alone, thanks to the unhealthy lifestyle of the modern society.
Having worked at Abbott Vascular, a major manufacturer of Drug Eluting Stents (DES), I’m very familiar with Drug Eluting Stents. Drug Eluting Stents (DES) is a combination of drug and devices but from a regulatory pathway, closer on the medical devices side.
The goal of this FDA project was to see if they can apply QbD methodology for a drug device combination.
Dr. Sharmista Chatterjee of FDA begins by explaining what a drug eluting stent is and then goes into how they implemented the QbD approach. Dr. Ken McDermott then explains in detail the experimental methodology, Design of Experiments (DoE) and the results. Finally the presentation concludes with lessons learned.
Enjoy this FDA case study on drug device combination.
This is going to be a Tag Team presentation between me and my colleague Ken McDermott. This actually started out as a research project. It is a research project between our office and Ken’s office. And what we wanted to explore in this project was to see if we can apply QbD methodology for a drug device combination. So that’s how it started.
So here is the outline of our talk. It’s going to be a tag team. I’m going to go first and then Ken would come in. Here is our outline we started by giving our primer on what is a drug eluting stent and talk about how we implemented the QbD approach. Ken is then going to come in to talk about the experimental methodology and the results and then the conclusion.
So, just to give a primary what is a drug eluting stent? These are implanted for treatment of arteriosclerosis. Usually what has been given in the past is that if you have these bare-metal stents that for almost 25% of the cases, they have cell growth on these bare-metal stents which then would call for need another intervention that is surgical intervention, which is not really desired. So now what the industry has been doing is to put these bare-metal stents with drugs like Everolimus or Sirolimus which prevent these cell growth.
And since the DES, these drug eluting stents, so in our study we were looking at these drug coated stents. The drug eluting stents (DES) status is stayed permanently in the patient’s artery that is very low tolerance for failure. These are made in very small batches, so it’s not possible to test each and every stent. So it’s very important that the quality is right.
Why we chose this was when we did a quick survey of what’s out there and what are some of the problems we’ve seen in our works in this field of drug eluting stents. This is what we saw for example, this is all from the internet from different publications.
One same manufacturing person’s complications for drug eluting stents and there was another one which says that there is a lot of problem with these stents and it says that this one has shown inconsistency in pharmaceutical composition. This company was saying that we’re experiencing issue with radiations in the drug release during the in vitro studies. Then there was another article from JNJ who is one of the biggest manufacturers for these drug coated stents for they have suffered a number of consumer product recalls because these stents were not functioning right.
And when we looked into these notes further and we looked at these articles further, we found out that the common quality issue is due to the high variability in the total drug content and the rate of the drug elution. And both these functions and both these attributes are actually related to the coating process of the drug eluting stents.
So why QbD? I don’t need to go into this in detail because in the past everyone has talked about that how QbD advocates use of science and facilitates consistent manufacturing and so we decided to take and apply the QbD paradigm to understand the ultrasonic spray coating process for these drug eluting stents.
So again, like all the previous speakers that we have seen, we followed the step by step QbD approach. We started on initially by defining or identifying both QTPPs and CQAs ready for these drug eluting stents.
Then we did an Ishikawa approach diagram to understand the factors which affected the drug elution. Next we did a bunch of DOEs, a step by step DOEs to understand the factors which were mostly important in our maintaining the high impact of the ultrasonic coating process and see if we can come up with a design space.
So like the slide here lists all the QTPPs and CQAs that we’ve identified. I’m not going to go into all of them but just to show once we’ve identified all the QTPPs and CQAs then we can show we’ve also established what is the link between the QTPPs and CQAs?
For example like potency and strength is linked to amount of drug, content uniformity, coating uniformity that maybe adequate insertion for example that is related to stent mechanical properties.
The next step what we did was we did a thorough risk assessment using this Ishikawa approach and again this was a huge team effort, we had a team from our office, ONDQA and from Ken’s office. And we looked at drug elution because we have to identify our front that the drug elution was the problem with these coated stents.
And we then went on to identify all the different factors which affect drug elution. So the colour coding actually shows the different type for example the ones in the pink like coating properties, drug substance, polymer those are the material properties. Analytical measurements are shown in green and then you have the manufacturing process for example the spraying, layering and then drying and the environment. And then we kind of move on to see what we can study within the scope of this one recent project and we identify spraying and layering was the 2 processes of this entire process being spraying and layering is what we were going to be looking at further.
Martin K. McDermott: Ok, so I’m really interested in medical devices and not QTPPs and all that stuff. But my interest is in how processing affects microstructure and how microstructure affects the properties of devices and their performance. So what we did, we made a solution of drug, polymer and solvent and spray these onto stents using a ultrasonic spray device.
And this ultrasonic spray coater is within this chamber which is a environmental control chamber. So basically what ultrasonic spray coater does, you take a coating solution and it flows onto a nozzle that is vibrating at high frequency and this forms droplets of the solution of the polymer and drug and solvent.
And then you have a flow of gas which focuses the droplets onto a rotating and translating stent. What we’re interested in understanding is how this process resolves inconsistent mass of coating on the stent. Therefore consistent amount of drops on the stent has released into the patient. So that’s one aspect of this. Others are these variables of temperature, power and flow rate of those solutions and rotation, frequency and those sort of things affect that. But from a materials science standpoint, I’m interested also in focusing on solvent evaporation from the link which that supplies to the stent because that determines the structure of the drop within the coating. Is it drop on the surface of the stent? Is it beneath these between the coating and the sub-stream or is it big particles of drop or small particle, that sort of thing affects the release rate into the patient. So, there are 2 aspects to it is how much ends up on the stent because not all the droplets are going to hit the stent and then how fast depending on your process will have different amounts of solvent evaporation. And that in turn will affect your growth structure which in turn will affect elution rate.
So the first thing we had to do with all these variables was eliminate defects. And we wanted to study how variation in these different variables affected ultimately the drug elution rate base on a mass of coating and structure. So you end up with defects such as fibres and agglomerates and webbing which we don’t want to see. And that’s due to too much flow onto the stent too fast and so the evaporation rate is not high enough. Now another problem was flow that was drying was too fast on the nozzle but spray droplets are coming off. And we form a precipitate which cause a droplet to form and this droplet will fall off onto the stent and that’s another type of defect.
So we first found what’s a range of values for these different parameters of processing parameters will result in no droplets to form and no fibres and agglomerates. And of course like everything in life is trade off, the higher the flow rate the more fibres we get the lower the flow rate the more drops we get.
So we had to do some preliminary experiments to optimize everything to get the first preliminary defects so we won’t have to worry about what effect these defects going to have on the elution rate. So that way we can just focus on this direct correlation between processing parameters and elution rate which we measure.
So our first DOE we selected these 6 processing parameters to stay based on a preliminary of test. We have a range of parameters and after that it was a screening DOE to try and understand what we’re up against.
And then to look at some data and see what kind of variation we get at elution rate and structure. And that’s of the coating. And then we eliminated 2 of these variables because we felt if they are going to give us as much information as these other ones as far as what effects these would have on the evaporation rate of the solution on the stent and of drug structure and that’s sort of thing.
And so we did these 2 DOEs. This gives you a sort of a pictorial idea of what kind of parameter values we use. We had within the THF acetone ratio in 3 different groups and this is our standard group and these are the 2 extremes. We also look at 3 different flow rates of this solution and that acts on this. On this side we look at the different distances between the stent and the nozzle, 3 different distances. And then the 4th variable is temperature shown in blue, red and green. And so we had a central point and then the extremes. So one parameter that we look at because typically, we want a smooth a coating as possible.
We studied how these different parameters affected the coating roughness. We rated these qualitatively of 1 through 4 but we correlated each of these numbers with we actually measured the roughness and correlated with these groupings.
This was a pictorial description of all the primaries we held the roughness greater roughness is these brown or grey areas and less roughness is green. We didn’t see a lot of statistical correlation with roughness. There are some trends. We saw some trends because for example, greater in distance that the droplets had to travel between the nozzle and stent, it tends to result in greater roughness probably because of greater drying of the droplets so it was less flow once the droplets hit the stent. But what we really focussed on is the drug elution rate.
This is just showing you for anybody that isn’t familiar with the different steps in this process. This is a time, each one of these is at different time points so you can put a stent and basically attest to. Then you fill it up with the elution media made a reciprocates up and down. You start at different time points and you take a sample which is analysed by HPLC for the drug content. And you refill it to and go back and go through the reciprocation again until the next time point. After the final time point of 48 hours we took the stent out and study the coating again by SEM and AFM.
And this is just to give you an idea of the 2nd DOE that we did. Of the types of elution canaries that we saw, this is on the Y axis, the Z accumulated drug release on the X axis is the time. And this is the centre point what you may say between the extremes. Yes, you can see I just point out that as you might expect when you vary these primer distances [00:16:08] looking at the different amounts of THF. We just give you an example how variation in THF because of some greater variation in the drug elution that we saw. And then this was for our first screening experiments and this is for our final optimization.
Now in order to relate drug elution within the primaries as an up, we’ll talk about a little bit we mathematically assign coefficients of this equation of each of those curves that I have just shown you, experimental data to this experimental equation in order to characterize the type release kinetics. For example when we came to coefficient for A1 relates to the initial first release amount, A2 which is a searching with time which is characteristic first release time. Then A3 is typical of long term release. So we can use these to try and understand the effect of the different processing parameters on the drug release. Is it the first or does it have an effect on the long term release?
Sharmista Chatterjee: So what we did next is we carry out as Ken separated these 2 of the DOEs. We analyse the data using Minitab and this did show us measure of the normality plots to figure out to which was a set. Again we look at a bunch of different responses like roughness and we look at those models that will fit in like A1, A2, A3 and we also look at amount of drugs that were admitted at 30 minutes and at 48 hours. So then what we did is from those contour plots are from those normality plots but we saw that 2 of the most significant parameters were temperature and the distance of the stent from the nozzle. So then we went onto come out or to develop or define these contour plots where on the Y axis this we have distance, on the X axis is temperature.
We look at the contour plots for 4 of the responses and then our goal was to see that if we can come up with a design space so this optimum region where we will minimize roughness, minimize burst release and maximize the long term release of the drug. Using this approach we saw that this is for example for a B which was a A3, this is the best region and if we overlap all of them we can put one overlay over all the 4 contour plots. This region comes up to be like the optimum region which will give us the desired goal.
And so from this we concluded that this was just a way to show that he can apply QbD methodology to this drug device combination and yet supplying us with a better enhancer understanding of this stent coating process. We figured out, we saw that nozzle to stent distance and temperature were identified to be critical parameters which affect the coating roughness and the drug elution kinetics.
We are also looking now to see if we can apply similar this step-by-step approach for other drug device combinations. When we did it quickly which is surveyed before we would even embark on this project and we saw that no far at least there isn’t any published literature which say that this kind of methodology has been applied to optimize coating for drug eluting stents.
Finally in acknowledgement, we have a lot of people to acknowledge. Senior colleague from our office, she is here. She was helping us with the risk assessment. Without the people in our office and the CDRH is truly a very collaborative effort. Funding for this project was provided by RSR grant and some from our own office and ONDQA Research Funds. Thank you very much.
MC : Thank you very much. Are there any questions?
Question 1: I was just curious because I saw when you spray that you kind of spray it on the outside, is it uniformly the coating on the inside versus the outside, the coating or the distribution of coating coz sometimes in contact with the blood vessels and some of the contact with the blood doesn’t that stuff matter?
Martin K. McDermott: The coating is primarily on the outside of course and that’s where you really want it. You want the drug to be absorbed into the tissue and not into the blood on the inside and so it doesn’t really matter. There is a coating on the inside but it’s kind of… so that’s fine.
MC: Any other any question?
Question 2: I have a question, if this is a question. If the coating on the outside is rough, does it have any effects on how the blood is going to accept it or anything like that or is this nothing, is just that the blood is so small amount that you don’t care?
Martin K. McDermott: We try to find a problem with the rough coating. Nobody could give us any reason to coating or not coating with their problem. I think it’s cosmetic anyway. I think it might, it does affect the resorption rate. These coatings are not resorbable so they go away eventually so a worry for coating will tend to tear aparts where the trough in the coating will go away quicker because it’s thinner. Other than that I don’t think that will have that much of an effect on the elution rate. So we didn’t see any effects and I don’t think there is that is very high.
Question: Presumably the rougher the larger the surface area so you’re talking about….
Martin K. McDermott: It’s not much greater surface area. It’s kind of like waves rather than like a roughness.
Question: I was going to ask one other question which is are you available for contract of service?
Martin K. McDermott: No.
MC: Any other any question?
Question 3: My question would be going forward like we talked earlier what is the plan for influencing devices industry to engage in more QbD aspects going forward. Any plans?
Sharmista Chatterjee: No. And this we have to first influence within the agency first and see. So this was just an exploratory project like you’ve said it was a research project was some small project that came to I with the colleagues we undertook. Next we plan if we have the results from this, we plan to share it in our office and with CDRH and see what the influence is. Now we’re taking it to the next step if we can talk to the device world that will be good but we’re just taking one step at a time.
MC: Any other comments? Alright, thank you.
Conclusion
We just heard from Dr. Martin K. McDermott of FDA/CDRH/OSEL/DCMS and Dr. Sharmista Chatterjee of FDA/CDER/OPS/ONDQA.
It is interesting to see FDA conducting QbD projects on combination products and devices. Although the QbD project was at the research level, I commend the direction that FDA is taking.