Transcript of conversation between Dr Ankur Mutreja and Dr Lara Marks, 2 November 2020

Ankur

I am Ankur Mutreja and I'm a microbiologist by background. And I've trained in industrial biotechnology after my basic degree, after which I got into public health, primarily from a cholera perspective. And when we were doing the Public Health Study of cholera, how it spreads, how the transmission patterns work. And we published our report, which was a global success, we were repeatedly hit by questions from all over the world, ‘how should we make use of this complex data?’ And that's when we decided that we need to simplify it and bring it to a level so that it's practical, and so that it is ready to then make a difference in the real world where these problems basically exist.

Lara Marks

I remember you telling me a story of how you used to go into the villages to collect samples and that it was desperation from the people in the village that sparked off your idea for a simple affordable diagnostic.

Ankur

Thanks for reminding me of that story. That takes me back to the time when we were studying patterns of global spread of cholera. And once we found out that the source of pandemic cholera is essentially the Bay of Bengal region, after that, we did a lot of work studying the Bay of Bengal region, as a sort of hub for cholera to get to the root cause of the problem. And when we were trying to study the source, as you would imagine, we were collecting samples from the region, we were visiting remote parts, very, very remote parts and villages, in that region, in parts of Bangladesh, in rural parts of India, where nobody understood English obviously and some of the local languages there were almost tribal.

So when we were collecting samples I was really taken aback by a comment some of the villagers made in one of the villages, and they said, 'you come here every year, you collect samples, you disturb the harmony of the of the village, and, and you go back with your samples, then we have no clue what you do with those samples. And if you collect those samples to make healthcare better here, or improve the landscape of whatever disease you are studying better for us, then we don't see that happening. Because it has not changed over the last few years anyway, since you've collected or you've been collecting samples'.

So that really shocked me... and made me realise that we've been collecting samples, we've been doing fantastic studies, that that of course needs to be done. But at the same time, we are actually not simplifying our output enough so that it becomes practically usable for ground level there the problem actually exists.

So that is when we decided that we're going to do something about it. We will make pieces of this genomic data. And from those pieces, we're going to try and extract very, very useful information and that useful information we can then use to create tools that could be used at the ground level.

But then you had, again, two choices: you can create complex tools, because it's a modern world, now, you have modern technology. But at the same time, the question lingered in my head, that even the tools that you design, or the tools in which you use your data, have to be simple enough, so that they can be used at the ground level, where these problems actually stem from.

So, we chose a clear path of simplicity at every single choice-making milestone that we had; whether it was choice of tool, whether it was how we simplify the data, how we can make the reading of the data, or the output of the tool, simple. And that's where we started this work on making portable diagnostics that would be field usable, and that would be easily communicable.

One of the important bits of public health is once you have a result you have to communicate that very, very quickly to the public health agencies so that action could be taken as per the results. So we worked on all these aspects. And here we are talking about portable AMR diagnostics, which is really a programme that is based on our previous programme of cholera detection.

Lara

How does your diagnostic work? How have you made it so simple?

Ankur

Genomic data means sequencing thousands of genomes. And then once you've sequenced thousands of genomes, that's actually the easy bit. The difficult bit is analysing those genomes, because you've got data from thousands of bacterial DNA, bacterial genomes, and that data could run into terabytes of data. And then you need tools to handle that data. You need tools to parse or read that data, you need a computational mechanism or computational power, that can actually handle that kind of data and then process it in the background and eventually give you an output that makes sense to us as scientists, as people who have sort of upgraded ourselves with modern technology.

But most of the outputs, we have to remember from genomics, actually if you put in front of a layperson, it would not make any sense. You know, those phylogenetic trees, those lineages and how the spread is working out, what is the transmission dynamics, what are the mutations, which are the changes, lineages and so on. I can go on talking jargon, but once you try and put that in front of a layperson it has to be simplified.

So when we looked at the problem, we thought that what are the questions that we're trying to answer from the genomic data. And then we simplified those questions as ‘yes’ or ‘no’ answers. So that if anybody wants to answer a particular question, they would have a choice of questions that they would be able to raise. And then based on those choice of questions, they can actually design the tool or they can customise the tool in their own way.

So the question that you have to answer for local improvement of a public health problem can be answered in the way they want the question to be answered. And that's how we came about designing the tool that would take genomic data, that would select the information automatically. And then the tool will use that information to design those probes. Nobody has to really understand what those probes mean. And, then those probes would get manufactured. And they would be ready to answer those very simple questions that we know people want answers to.

Lara

So the healthcare worker gets the test and they have a sample - how does the diagnostic work and how do they read the results?

Ankur

Yes, at the moment, we're trying our best to make the test work directly on the clinical sample. We've done a lot of testing in-house where our, our, our, if you want to call it kit or a detection test at the moment, it works beautifully on ... total nucleic acid extracted from any clinical sample. But the end goal is to make it work directly on a clinical sample.

So then to answer your question - our end point is that when we collect a sample, whether that is a stool sample or it is a cerebral spinal fluid (CSF) sample, it is a blood sample, a urine sample, or any particular swab from from your throat, or vaginal swab for a UTI infection, for example. In that case, you would directly be able to use that sample in diluted form as a direct starter for the test. You would have thermo stable reagents provided as part of the kit and the sample that you collect from the patient will directly go into this test, it will actually resuspend those thermal stable reagents and be ready for putting in into the PCR machine that is again provided in the in the suitcase that we've designed from a portability perspective. And then once you've you've hit the run button on the machine and the test is finished, you just take a paper stick which is again provided in the assembled kit, and you dip it in the tube in which you had all the reagents, and you would have the result in front of your eyes.

And we've actually gone beyond that. So instead of just seeing the result in front of your eyes, we are actually designing a mobile phone application to avoid any room of error. That mobile phone application will read the test that you will generally read by your eyes and then note the results. And we don't stop there either. Then the result that the mobile phone or tablet has recorded, with its camera, is, after it is recorded and saved, it is ready for communication as well as using any form of internet.

Lara

Okay, so it can be uploaded to a central database.

Ankur

Yes.

Lara

So if you're in the clinic and you're a doctor, you take the sample and then what happens to that sample? Do you put it into a PCR?

Ankur

So that sample, you put it into a PCR machine, which is provided as part of this package. And once that PCR has finished running, you read the result on that stick as I previously mentioned.

Lara

How long does the test take to run?

Ankur

So PCR is pretty standard, it takes about two hours to run. In some cases, the results are ready in 90 minutes. But there is some pre-processing that may be needed once we're going to optimise our tests to work directly on clinical samples. That procedure might take an hour or so, which is a pre PCR procedure. So overall, the readout of the result, once the PCR is finished, is only going to take 10 to 12 minutes.

So we're estimating that the whole process of processing the clinical sample and making it ready for a PCR and then running a PCR and then reading the result, and then recording the result on the mobile phone will take somewhere between 2 to 4 hours.

Lara

And going forward. How do you see this being developed? How long do you think it would take to roll out?

Ankur

In terms of rolling out, there are two key questions. One is we are trying to optimise it, so that it works directly on clinical samples. That work might take six months to a year, depending on what is the question that you're trying to answer. Some of the clinical samples are easy to optimise, and the others are more complex. Stool as you can imagine, is a lot more complex than you using a cerebral spinal fluid material, which is much cleaner.

And at the same time, if there is a case of sepsis, for example, blood is also relatively cleaner to be used as a direct starter material for a PCR test. But a stool needs a lot more optimisation, because it has a lot more contamination, it has some material in it that can actually block the PCR from running as well. So you have to dilute it. So all of that optimisation work takes a little bit longer.

We estimate that once we've optimised it in the next six to 12 months, we're going to put this to test in a real world setting, which would be in various places in India and Africa. And in some cases, when we talk of AMR in particular, the test is applicable also for developed country settings. We're going to put that to test in a study and try to establish a statistical significance by meeting a set number of samples and come up with a test which is absolutely robust. So overall, we're estimating the whole thing to be ready in two years time for rollout.

Lara

And when you're testing, are you testing to identify the pathogen or are you testing for something else?

Ankur

You could do both. You could identify pathogens, so the probes that you have in this case could be made such that they can actually start with identifying the pathogen and then they can identify the resistance. But in many cases treatment can get delayed, if you're trying to establish what pathogen it is. By looking at the symptoms, a doctor can generally tell what kind of infection that may be. So it's much better to establish what the resistance portfolio of that condition is. So that an empirical definition, based on the clinical symptoms, and the resistance portfolio based on this diagnostic method.

Lara

Can you explain what a resistance portfolio is.

Ankur

So when I say resistance portfolio, that means what exactly the condition might be resistant to. And when I say resistant to it, I'm talking about resistance in terms of antibiotics. When doctors are prescribing antibiotics, in most cases they rely on the current public health landscape, as it is known to a doctor. But public health landscape records are not updated in real time. And they can often be wrong in making decisions. So therefore, a true picture, or a true portfolio of the condition in terms of resistance, can only come from its diagnosis there and then. And this is what we are aiming to provide.

Lara

That presumably is based on a lot of genomic data in the background that's been compiled.

Ankur

Absolutely. So we've mined thousands and thousands of genomes. We've mined databases that have been put together specifically for resistance genes. And we've gone into very deep depths as a fishing exercise, and then we've come up with these probes.

Lara

And what were the challenges in coming up with the probes.

Ankur

The main challenge is that what we've designed, for example, is 16 probes. But the problem at hand was detecting over 100 different variants. In a normal world, you would have one probe designed for detecting one type or one variant. So therefore, for any particular problem, you would have to put it to test, in a system that can give over 100 answers. Whereas in our case, what we've really done is do all the informatics and all the smart work of designing probes in such a way that with only 16 answers, you can detect more than hundred different types of resistance. And that's where the unique selling point (USP) is. And that's where the technical complexity of probe designing was.

Lara

Is there anything else you would like to add that you feel we've left out?

Ankur

I think the main thing is the problem. AMR is something that has been increasing and increasingly ignored. People are paying attention to it now. But I don't blame anyone for ignoring AMR because the right tool that is needed for detecting and decision making in AMR does not exist at the moment. So there is a huge need for a proper testing system to be in place for proper diagnosis of AMR that has to be put in place. And we’re just starting in terms of diagnosing AM. Therefore these are early days in the AMR detection system.

Lara

What is the real issue in poorer countries? Is it different from what we experience in the more wealthy countries?

Ankur

One of the main differences is that the prescription of antibiotics in poorer countries or developing countries is not really required for buying an antibiotic. And that, again, is no one's fault. It's mainly because the infrastructure there is not capable of having a prescription based model at every single for every single patient. Because the burden of disease or the burden of different conditions, is very different in poorer countries as compared to these countries.

So if every doctor was going to prescribe an antibiotic based on proper diagnosis, which at the moment could take days, then a lot of people would die just in that time. So, at the moment, the balance is being struck between care and decision making. And this is the balance that we're trying to maintain, but at the same time make clinical decision making better in poorer countries.

Lara

And for the over-the-counter antibiotics, how would this diagnostic help in that situation? I can understand how it would help within the clinic setting but what about at the pharmacists and buying an antibiotic?

Ankur

Every country essentially has rules in their rulebook that over-the-counter sale of antibiotics is prohibited. Whether you talk about any country in Africa or Asia, where we hear about over-the-counter sale of antibiotics, their rulebook actually says that no antibiotics should be sold without prescription. And every antibiotic, a strip of antibiotic tablets that you buy in those countries has a marking which clearly indicates that this is a prescription only drug.

But again, as I said, this is a balance between provision of care and decision making. And to keep provision of care at front, as it has to be, over-the-counter, sales of antibiotics have to be basically ignored, because if they are not ignored and they are strictly enforced, then there will be a catastrophe.

Lara

So how do you think your test could fit into that situation?

Ankur

So that's exactly what we're trying to fix. The main reason why a prescription based model is not applicable at the moment for buying end devices from a pharmacy is because there is no proper diagnostic system available. If there was a proper diagnostic system available at a local GP or at a local roadside clinic, then they'd be able to subject the patient sample to test in house, locally on the site, and give a proper decision on what antibiotic should be given. And once that is in place, then the government would have to enforce the rule better and ensure that the wrong antibiotics are not sold, for problems that will not be treated with those antibiotics.

Lara

Okay. And can the test be run without electricity and without big laboratory equipment?

Ankur

The USP of the test is that everything comes in a suitcase, which can actually be transported as a whole suitcase, it could be carried in a plane, it could be carried to any place, it doesn't matter how rural it is, it does not require electricity to run. But if there was electricity, it can be plugged into an electricity socket. But it could run using a battery, which could be a car battery, or a motorbike battery, if that was available. But to supersede everything, what we've done is we've also built in a solar panel in our suitcase that would provide continuous charging to a battery that will come with the case. And that battery can then just run the PCR machine.

Lara

What about the solar paneled battery you mentioned?

Ankur

So the solar panel is going to be included in this and that solar panel will automatically charge the battery. And that battery could then be used for any purpose whether that is to charge the mobile phone or the tablet device that comes with the package, and also to run the PCR machine that is actually carrying out the test.

Lara

And one last thing because I know your time is running on. You talked about the issue about the reagents used in the kit and the difficulties of keeping temperature control. Can you talk a bit about that.

Ankur

Yeah, so that’s another great question. Because when we were trying to simplify things, we were hitting hurdles at every single point because it turns out that actually simplifying things is more complex than any complex thing. So when we were making portable kits, a question came to us that, you know, ‘you're claiming that you would use your kit in a very rural setting, for example, in Uganda where temperatures can reach up to 45 degrees celsius and most of the molecular reagents that are used in the lab, are quite sensitive to heat, and they could really go bad pretty quickly’. So in answer to that, we had to do a totally different manufacturing exercise. So we worked with a PCR, reagent manufacturer, and we have given them the recipe if you like of our test with all the reagents and and they've gone ahead and made the lyophilised form of those agents. And by lyophilisation, what it does is that it gets rid of all the water that there is in the reagents, and it leaves it into a powder form, which is highly stable at very high temperatures for long periods of time. And that's how we've been trying to answer the thermal stability or the thermal sensitivity problem.

Lara

All right. Well, I appreciate your time. Thank you very much for talking to me.

<< Simple, cheap AMR diagostic at the point of care

Interview with Archana Madhav >>