0:00
Hi, I'm Matt Pierce and I'm going to talk to you about productizing your diagnostic.
0:08
Hopefully I've pitched this at the right level, so it might be introductory level or it might be and maybe a bit too simple.
0:15
But if not, you can feel free to approach me afterwards.
0:19
But basically just a bit about me just quickly.
0:21
So I've been in the diagnostics industry for about 35 years now, having worked for some of the big national multinationals.
0:28
But also, I run a company called Medtechtomarket.
0:31
And we're a contract design, development and manufacturing organisation taking products all the way through from, you know, translational biomarkers all the way through to commercial launch.
0:44
So I'm going to talk a bit about the productisation process and what's it matter.
0:49
So we see this a lot, but great science does not always mean a marketable product.
0:55
And the challenge is really to productize that technology into a marketable and clinically relevant product.
1:06
So the main focus of this presentation is in that development pipeline.
1:10
And I'm really looking at this sort of design development, the manufacturing validation and verification and validation processes and what's involved in turning a biomarker into a diagnostic test and diagnostic system, the regulatory submission work and all that will follow all of that.
1:26
And I'll touch upon some of that as well.
1:30
So I'd like to use the analogy of baking a cake.
1:32
The diagnostics to developing a diagnostics product is like baking a cake in the sense that the process of making the cake is pretty consistent and you follow the steps 1 by 1 and you always bake at the end before you mix first.
1:47
But the actual ingredients that go into that cake dramatically change.
1:51
And so this sort of presentation is going to talk a bit about the ingredients and a bit about the process and trying to sort of harmonise those two together.
2:00
Just a couple of things, essentials just in case there's, if there's industry colleagues here, then you'll probably know a lot of this already, academic colleagues, possibly not so much.
2:09
But there are essentials that you need to do in order to submit a technical file or a product and become a legal manufacturer.
2:16
The first one is you need a quality management system in order to be a legal manufacturer.
2:21
And a legal manufacturer means it's somebody who owns the product design and is doing the registration.
2:26
And that really that quality management system refers to all the procedures of how you go about your business and the certification of your business.
2:35
But within a product and a product development, there are three or four or five essential ISO standards, which you also need to be compliant with to do with managing risk, having a usability engineering evaluations done and all of your labelling and your instructions for use also have to comply.
2:54
So those are the kind of essential elements.
2:56
That's all I'm really going to talk about for regulations today because we could get go down a rabbit hole on that.
3:01
But pretty much everything from now onwards is all about understanding your labelling claims and what you're going to need to do in order to productize your IVD.
3:10
So the first thing you have to consider is where is the intended use statement, intended use, where your products used, who by how, when, where all of those elements and what for and also any exclusion limitations or exemptions.
3:26
Now I'm not going to go into the detail of this slide.
3:28
I know it's a busy slide, but basically I'll put here just as to indicate what why it's important in IVD is you have here's are 2 glucometers effectively the same technology that's gone into both of those two tests.
3:40
They're virtually identical in terms of their performance.
3:44
The one on the left where I've highlighted in the colours based on of the intended use statement of what it's used for, where it's used and everything else.
3:53
And if for example the patient type, I've highlighted those.
3:57
But this is a glucose test.
3:59
It's pretty standard stuff.
4:01
It's for useful people with diabetes and then where it's meant to be used at the bottom line there it says the system is meant to be used by a single patient and should not be shared.
4:10
That's a really important statement that's been put in there has been forced upon them to put that in there, but it's been put in there because their validation is dictated to a product development journey is going to be dictated by whether they can support that labelling claim.
4:27
On the other side.
4:28
The ACCU CHEK is they talk about, it's used in intended for multiple patient use in a more professional setting and that will drive a whole new change of validation.
4:41
So I can't stress this enough.
4:43
If you're starting, you've got biomarkers that you're looking to develop is you have to start with the intended purpose and you have to get this piece right because this will drive everything in your technical file submission.
4:57
And you know, so this will dictate all the product design development and everything there.
5:03
So how do you start with the intended use statement?
5:06
So this is the basis of your labelling claims.
5:09
So you need to think about market research, clinical engagement, talking to clinicians.
5:13
How is this going to be used, where it's going to be used to really finesse your intended use statement for this audience?
5:20
There's also possibly a companion diagnostics element you need to consider.
5:24
And for that, the best place to go is your scientific validity.
5:27
So start with scientific validity there and think about where is the evidence for this technology and how it's going to be used because the wording of that can then go into your intended use statement.
5:40
Once you've got your intended use statement, I'm whizzing through this step by step.
5:44
You're then looking at design inputs.
5:47
Now when you do a technical file submission for whether it's FDA or European notified body, yeah, notified bodies, you will be submitting what they call a traceability matrix, which is effectively a design inputs documents.
6:01
That's your design specifications.
6:03
And you will also, I mentioned to you about risk assessments that you'll also be doing risks.
6:07
I'll come to risk in a minute.
6:09
So you need to consider based on that intended use statement of what you're doing, of where your product is going to be used and who and how and when and all the rest of it.
6:17
You also need to consider then what are all the design features that need to go into that technology, into that product, I mean, in order to meet that intended use statement.
6:28
So you're considering things like, you know, safety, it might be electrical safety if it's an instrument or radiated immunity testing or whatever, but also the fundamental assay performance.
6:38
Is it a clinical chemistry application?
6:40
Are you talking about sensitivity and specificity?
6:43
You know, all those different aspects of how the test will be used and then starting to draw out.
6:49
So this will be, I mean, this is going to be pages of specifications including acceptance criteria.
6:56
Now this is design intent and traceability are really important with the regulators.
7:00
When they do the technical file review, they will be looking at this and they'll be looking your design inputs and your design outputs.
7:08
So it's really important to get this piece right at the beginning.
7:11
Now, in some cases you can kind of retrospectively fit this if you've already started your project without having some of this stuff in place.
7:18
But I can't stress enough, get the intended use statement and get the design inputs and design specifications sorted early.
7:27
Again, I'm not going to go into the detail the but risk management is also critical.
7:32
You're going to have to submit that as part of your technical file and it's ISO14971.
7:38
Basically, risk management is a process that you're going to follow throughout the whole of the project.
7:44
So you start with a risk management plan and that's really classifying how severe your product is.
7:50
You'll have done a regulatory assessment as well, but how significant the risks are to your product and the likelihood of those events happening, but also quantifying how you're assessing those.
8:02
And then throughout the project, you revisit the design specifications.
8:06
Oh, sorry.
8:07
The risk assessments, because if you identify risks that most risks need to be designed away if you can, and those also form part of your design input.
8:18
So now if you do an FDA510K submission, for example, they will ask for your risk management plan and your risk assessment, and you will need to provide that information with them.
8:27
And you will revisit this iteratively throughout the project.
8:30
So every time you do a design change, you should be revisiting the risk assessment to check that there's been no new risks identified or generated.
8:40
Now we're into design.
8:41
So at this point this is where the ingredients really matter.
8:44
So is your product a lateral flow test or is it a PCR kit or is it an immunohistochemistry reagent pack?
8:53
Whatever those products are, you're then starting to do the development.
8:58
So you're looking at performance evaluation plan.
9:00
That's the first thing you should be writing is to work out how you're going to qualify the test.
9:05
Now that should include analytical performance, also your clinical evaluation planning and also if it's companion diagnostic, you also need to start considering how this is going to be presented to the European Medicals Agency.
9:19
So and scientific validity.
9:22
I will say at this point, start planning your clinical trials because they take time as everybody will know.
9:27
But really the chemistry and any engineering work that's doing is you're starting to develop this.
9:33
So starting to optimise the chemistry of your assay because as I'll come to in a minute, getting to the product as quickly as possible is really important.
9:44
Carrying on as you get to the design.
9:46
So you've started to optimise the chemistry now you start to integrate it, whether it's a diagnostic system, it might be an instrument and a disposal.
9:52
If it's point of care testing, for example, it might be an instrument and disposable tests.
9:57
Or if it's, you know, a Multiplex PCR kit, for example, you're starting to compile and build the data to show that the Multiplex is performing and everything else.
10:09
So you start to pull together and then maybe interpretive software algorithms and applications that you need to develop alongside.
10:17
Remember, you also need calibrators on a lot of these products in order to submit them.
10:21
So you may need calibrate or controls in order to demonstrate performance.
10:26
Once you finish the design at this point, you've now got a fully integrated system that's effectively designed and you can consider it design frozen.
10:37
And then at that point, before you really start your clinical validation and your analytical validation, we would suggest, we always recommend that you do the manufacturing validation piece now, because then all of your clinical performance and your verification and validation work is demonstrated on the product that the customer is going to buy.
11:03
So, and that's so from that moment onwards, once you've got to design freeze, you can actually, you're doing all of your technical file based on the product that you're manufacturing and you have lot history and batch history based on that.
11:16
So the manufacturing validation will be setting up all the various procedures, SOPSQC processes, risk assessments for your process, all those kind of things, master validation plan, all of that work needs to be set up as part of your manufacturing.
11:32
Once we once you have to manufacture product, then you're into both clinical and analytical validation and verification.
11:40
I've only got 20 minutes.
11:41
So, but this will be a list of things like stability studies, limits of detection, linearity, precision, you know, depending on your intended use.
11:50
We talked about, I talked about, I should point to that intended use setting of whether it's the instrument, the glucose metre can be reused on multiple patients.
11:59
That's really important because if it's used on multiple patients, then you've got infection control to consider between a bit on that system and so that device will need to be cleaned and validated on a decontamination validation or its whole equivalent of its life cycle of the product.
12:17
So that might be 10,000 cleans you need to validate.
12:21
So that's why the intended use statement will dictate a lot of these different things, distribution and robustness distribute this.
12:29
There's regulatory standards for all of these things to give you guiding principles.
12:32
But basically distribution depending on if it's point of care near patient, the distribution of that product will be very different compared to if it's going to a hospital lab, for example.
12:44
And so, and the same with things like drop testing.
12:47
So all of that work will feature as past your building up your technical file. In parallel, you can be doing your you can be doing your clinical evaluation, clinical validation.
12:58
Now this really does depend on the product type.
13:02
So there's a whole some being fairly generic in this presentation, as you can see.
13:06
But basically the first thing you need to consider is your intended use setting.
13:10
So is this going to be a home use test?
13:11
Is this going to be in a clinical lab?
13:13
Is it going to be near patient?
13:15
So those elements are of you need to demonstrate within your clinical process.
13:19
Now I know we're in a precision medicine conference.
13:23
So the clinical chemistry may not be quite as relevant, but clinical chemistry.
13:28
So typically you're there looking at method comparison assays.
13:32
So you're looking at the bias and precision bias and yeah, accuracy against predicate devices, which might be a lab reference method.
13:42
And those kind of studies actually aren't particularly punitive.
13:45
So 120 patients will be sufficient to demonstrate your concordance with the method comparison.
13:56
And for European submissions and UK obviously you can get away with one site.
14:02
So those studies are really quite achievable within the within Europe.
14:08
The US typically would have three sites, but it's still 120 patients. In terms of oncology and some of the other tests then really and companion diagnostics fall into this.
14:21
Then diagnostic performance actually becomes a slightly bigger study.
14:25
So typically they're powered by the prevalence and the confidence intervals that you need to work around.
14:31
And they typically between 4 and 500 patients.
14:33
So nowhere near as bad as phase three clinical trial or anything like this.
14:38
But so you can still drop these in and you can use biobank samples in a lot of these studies.
14:45
But again, for a companion diagnostic, it's Class C you will need to engage with FDA to get you what you…, I would advise to do a Q sub with FDA before you start at these projects and finalise your clinical trials to demonstrate out that they're going to be happy with your study design.
15:05
Some projects you also need a human normal reference intervals.
15:09
So depending on the biomarker of choice, whether you need normal patients or healthy patient comparators, there's also precision studies.
15:21
These can be split into two types of precision studies as the CLSI standards will dictate some of this stuff.
15:26
But basically the precision studies typically are a 20 day study done by the manufacturer at the manufacturer's site and that's 20 days multiple operators, multiple lots, hence the manufacturing validation using different high and low or thresholds and usually on the matrix that you're measuring against.
15:46
You can also do five-day studies, typically three of them in the intended use setting.
15:52
They're a lot easier to achieve on an individual basis.
15:55
But obviously you're outsourcing that to a clinical site, and you may need things like whole blood precision or urine precision studies.
16:05
So you're actually doing precision studies on the matrices themselves.
16:08
And then you finally with companion diagnostics, specifically you do also need to make sure that your clinical trial is powered based on the prevalence, but it's also supporting how the CE mark is going to be used and how the product is going to be used in informing the prescription, whether it's, you know, stratifying the patients as suitability as potential responders, or whether it's look at measuring the actual response to the therapeutic.
16:38
So your study does need to design in an element of being able to monitor those things.
16:45
Just a couple of other things.
16:46
So software is software as a medical device is massively increasing and we're seeing a lot of that.
16:54
And that again, you obviously need the validation cohort separate to your training cohorts.
17:02
So that data needs to be generated from virgin patient data.
17:05
And in terms of the attended use, coming back to the intended use statement, you may also need to do multiple comparisons.
17:12
So it might be the if it's a self-test product or a home use product.
17:17
You'll need a professional to compare to the naive user so that you're demonstrating analytical performance based on those two compared to the lab reference method.
17:29
You can use biobanks as I mentioned before.
17:32
And then some of those biobank samples can be used back up by the manufacturer at the manufacturer's site.
17:38
So you can collect the samples and bring them in, provided they're processed and collected in an intended use setting and a mechanism.
17:46
So, you know, if it's an RNA test, it needs to be processed, the extraction needs to be processed appropriately.
17:55
Labelling, instructions for use and labelling are very important.
17:59
I always think to recommend people start this as early as possible as well.
18:06
For some bizarre reason, it takes forever to do this stuff.
18:10
But basically you need to consider your operating manuals and your instructions for use.
18:16
Obviously, you know, I don't really need to go into too much detail on that.
18:21
But again, you will be submitting this part of your submission.
18:25
So and part of that submission really is the technical file is effectively as written there 1 to 9.
18:30
So that's effectively what you're going to be submitting for IVDR, I mean USFDA is actually very similar kind of structure of what you're going to be thinking.
18:40
So risk assessment, risk analysis, risk management plans.
18:44
And just as a rule of thumb, in order to develop a typical diagnostic, doesn't matter whether it's a lateral flow test or a more complicated instrument and disposable test cartridge, typically to build a technical file, you're running between 40 and 50,000 tests.
19:00
That's our experience of it, to do all the stability studies, precision studies, all that analytical validation.
19:06
So it's a big commitment to have all that data, but you do need to retain traceability because it will be audited.
19:14
So and that means you know, every operator test result needs to be documented and retained and being able to you need to be able to call that data up.
19:24
Thank you.