NEB Podcast #49 -
Interview with Mihir Kekre: Antimicrobial Resistance and Pathogen Surveillance

< Return to NEB Podcast Home

 

Transcript

Interviewers: Lydia Morrison, Marketing Communications Writer & Podcast Host, New England Biolabs, Inc. & Elizabeth 'Betsy' Young, Product Manager, New England Biolabs, Inc.
Interviewees: Mihir Kekre, Commercial Partnerships Manager, Tropic Biosciences


Lydia Morrison:
Thanks for joining us for this episode of the Lessons from Lab and Life Podcast, brought to you by New England Biolabs. I am your host Lydia Morrison, and I hope this podcast offers you some new perspective. Today I am joined by Mihir Kekre, who spent 4 years at the Centre for Genomic Pathogen Surveillance at the Wellcome Sanger Institute, and who shares his experience in antimicrobial resistance research and pathogen surveillance.

Mihir Kekre:
Thanks, Lydia. It's great to be here, and thanks for having me on.

Lydia Morrison:
Of course. My pleasure. And I'm also joined today by my colleague, Betsy Young. Betsy, thanks for being here as well.

Betsy Young:
Happy to join you.

Lydia Morrison:
So we're just going to dive right into the questions. I was hoping, Mihir, that you could give us a quick primer on antimicrobial resistance.

Mihir Kekre:
Sure. So let me start with antimicrobials first. So antimicrobials are drugs that are used to kill disease causing microbes. The most common class, as we all know, are antibiotics used to kill off bacterial infections. But AMR, or antimicrobial resistance, is when said microbes evolve or mutate over time and in ways that help them withstand the neutralizing effects of the drug, thereby rendering it ineffective, so this basically leads to deaths that are otherwise preventable. Now, AMR does occur naturally over time in the ecosystem with genetic mutations within a pathogen just strengthening its guard over time, but it can be accelerated and that effect is manmade, so poor infection control, limited access to medicines, and most importantly, the lack of public awareness and consequences, because most countries tend to use, overuse rather, antimicrobials a lot. So that's what generally causes AMR to accelerate over time.

Lydia Morrison:
And why is it important that it be monitored?

Mihir Kekre:
So let's just take the last three years, we've lived through some of the scariest times in modern history facing a pandemic. So words like pandemic, disease tracing, social distancing has all become household terms now, but what we've failed to take stock of with acute global disasters like COVID-19 is this background level of silent pandemic, as I like to call it, called AMR, which is barely above the surface, you never really see it happening and has devastating effects globally. So just to throw some facts at you, AMR was responsible for 4.8 million deaths globally, and recognized threats like HIV or malaria are just over a million combined, so you can see what a credible threat AMR really is.
But I guess the primary reason to monitor AMR is more the cost that it bears to the economy. So it's not just illness and death we're talking about here. It creates a significant dent on the economic circumstance of the country, so prolonged hospital stays, increased expenditure on developing drugs by the government, or just disruption to labor, because they get sick, people have time off work, so there's a huge commercial impact, and it definitely needs to be monitored.

Betsy Young:
Thank you for that background. It was very helpful. Now for maybe a little of your background. What is the mission of the Center for Pathogen Surveillance, and what was your role there?

Mihir Kekre:
The Center for Genomic Pathogen Surveillance, or CGPS as they like to call, it's based in Oxford, so it's headed by Professor David Aanensen with a goal of creating global AMR eradication reach through local impact. So at its core, the center builds software surveillance tools that are designed in-house to facilitate rapid insights using genomic technology. The focus basically is we make robust decisions with the data in front of us really quickly. Many of the tools that have been developed in house are endorsed by governmental agencies like the CDC, the European CDC, or even the United Nations. My role here was genomics operations lead, so it was more just to bring this added dimension to the center, adding a physical genomics footprint in the form of in-house laboratory capabilities. It was mainly to oversee and scale the centers sequencing operations right from sample and data collection through to generation of whole genomes across all of our international partnerships.

Betsy Young:
That's great. Thank you. As we saw with the pandemic, obviously scaling was critical to our response. You've also worked at the NIHR Global Health Research Unit for Genomic Surveillance of Antimicrobial Resistance. What work did you do there?

Mihir Kekre:
So the GHRUAMR was a large initiative as part of the CGPS, or the center, so it's a well established fact that low and middle income countries with rudimentary public health systems like clinics and diagnostics tend to bear the biggest and greatest impacts of AMR. So the Global Health Research Unit, or GHRU for short, was set up in 2017 with a goal to transform how AMR surveillance is done specifically in low and middle income countries using tools like whole genome sequencing. So currently there are well established surveillance systems around the world, but they're mostly limited to developed economies.

Up and coming region, well, they just lack the sort of comprehensive monitoring we want to get to. So what we did at the GHRU was to strategically partner with four global sites, one in Columbia, another in India, in Nigeria, and in the Philippines to make sure that we deliver AMR surveillance locally on a very, very holistic level. And these were diverse sites, some of them were national laboratories, others were academic institutions, some of them were even enterprises, and they all had different levels of AMR expertise. So eventually after four years of the project we could see a lot of retrospective cross-learning between sites, and that was one of the main reasons why we were so successful with the implementation.

Betsy Young:
That sounds like a Herculean task. At a very basic level, what is a typical AMR monitoring workflow? How are samples collected, how are they processed, and how is the data analyzed?

Mihir Kekre:
Sure, that's a great question. So within a typical AMR sample workflow, samples are collected from patients, or you could even collect them from animals, from farms, from the environment, and some basic tests are run to actually identify the pathogen on site, so at the collection site itself. There's also a bit of metadata that's collected with each sample, sample attributes like unique ID, patient details, diagnosis, maybe the location at which the sample was collected, date of collection and so on. I'll explain in a minute why these attributes are so important. The sample is then transferred to a more established laboratory where AST, or antimicrobial susceptibility testing is performed. So basically this means that the sample is put through a battery of tests that are mostly visual at this point, hence we call it phenotypic characterization. So they're grown on Petri dishes in a special growth medium, they're exposed to a panel of antibiotics with the sole goal of determining at what strength of drug they stop growing.

These results are then referenced against a clear set of controls to conclude whether the isolated is susceptible, or resistant, or even ultra resistant. It just gives you a measure of the resilience of that pathogen to an intense drug pressure. And then coming back to my point around sample attributes, we then use these inferences around drug resistance and then superimpose them on geolocation, and date, and patient details so that we can establish a background as to how these bugs are circulating in the mass population. And then following that, we could even make links between the various streams. And if they suggest a pattern, then we can try and map out if that shows us a disease outbreak.

Lydia Morrison:
And how do you link those various strains?

Mihir Kekre:
So we generally have software to actually link the strains using the metadata we've collected, and they are characterized based both by the virulent or the resistance they've shown on those battery of tests, but also by how their genome looks, so each pathogen or each microbe has a different genetic signature, and it's these unique signatures that we look for when we're trying to actually link lineages, and try and work out if they contributed to an outbreak or an epidemic.

Lydia Morrison:
So what's the value or benefit of augmenting the existing workflow with a molecular tool like whole genome sequencing?

Mihir Kekre:
AST, as I described, are the traditional testing methods and they usually form the cornerstone of most surveillance plans everywhere, they are pretty much gold standard, but they do come with certain limitations. For one, they have poor scalability. You can't really scale them across a national level program because they are time consuming, they're labor intensive, and sometimes take days to generate results. Also, every different pathogen or bug that you test for has a separate distinct laboratory workflow. So you've got to have a lot of staff, you've got to have them trained in specific testing methodologies for different microbes, so it doesn't become standardized as such, and sometimes you have scarcity of that domain expertise, especially in low and middle income countries. So here is where molecular assays like whole genome sequencing really add a new dimension, because they provide a lot deeper resolution to analyzing these pathogenic outbreaks, but they're also standardized, easy to do, and can be scaled to a much higher level than traditional AST.

And added to that, once you sequence the genomes, they can be retrospectively and repeatedly looked at over time because they're basically banked as data. So it's a permanent archive that you can go back to as many times as you want, which you cannot do with traditional methods. So it's crucial to point out here, just to temper this a little bit, that for actual clinical treatment in the future, we always try and recommend sequencing being used in conjunction with these traditional methods, because they only work really well when they compliment each other rather than being used standalone. But having said that, sequencing does offer some added advantages from a scalability perspective.

Lydia Morrison:
It sounds like sequencing can be a really powerful tool to help the monitoring process. I'm curious which pathogens are worth monitoring and how do you choose those pathogens, and are there pathogens that we should be monitoring that we're not?

Mihir Kekre:
That's a very good question. Out of the four plus million deaths that I mentioned occurred due to AMR, about 1.25 million are actually bacterial related, so there's definitely scope to choose your bugs carefully. There was a study done I think that surveyed over 200 countries worldwide and it estimated there are 88 different bug drug combinations that can lead to AMR. Just think about that. That's 88 different avenues that we have to tackle all at once. So as part of the GRU initiative, our focus was to profile a special set of seven out of these, commonly referred to by the World Health Organization as the priority pathogens.

Now they call them that... Well, actually I should say this is mostly a list of 12, but we included the most critical seven bacteria most commonly associated with hospital and community infections. So this list comprised your usual suspects like E. coli, Staphylococcus aureus, otherwise known as MRSA in common terminology, Klebsiella pneumoniae, Pseudomonas aeruginosa, and other bugs that cause tuberculosis and pneumonia as well. Now, out of the seven, the most alarming fact was that only two out of the seven actually have focused global intervention programs designed around them. So the top five out of the seven have no real sustained global effort. And this is what we at the GHRU wanted to address.

Betsy Young:
Can you tell us a little bit more about how the GHRU set out to improve the existing surveillance model?

Mihir Kekre:
Firstly, I guess we introduced scalable and sustainable utilization of whole genome sequencing as we've been talking about, and it's been a valuable supplement to the existing surveillance system. So as sequencing costs start to drop, we'll see an increasing number of countries trying to incorporate WGS as part of their national programs. Also, this improved surveillance model with sequencing can help reduce the use of frontline drugs that are becoming more and more ineffective, so you can try and use narrow spectrum drugs, as they call it, to target specific pathogens and just try and blasting any species in question.

Now, specifically for the GHRU, I think working with the four national laboratories we really were able to demonstrate the immediate value of genomics and how it can help with tracking high risk pathogens. For example, in just the first four years of the project, we had several success stories. For instance, the team in Nigeria was able to show with next NextGen sequencing that they could identify an outbreak of E. coli within a hospital ICU and then support the underfunded hospital to manage the infection. So this was just a very quick turnaround within a matter of weeks that they were able to identify, manage, and kill off the infections. So this is just one example just to demonstrate what the impact our surveillance model has actually had.

Lydia Morrison:
I think that's a powerful example though.

Betsy Young:
Absolutely. And earlier you mentioned holistic surveillance, what were some practical considerations that GHRU has taken to building holistic surveillance laboratories in those low and middle income countries that you mentioned?

Mihir Kekre:
There were several considerations, but I think if I had to break them down, there were four main blocks to this initiative. So one was obviously, as you can imagine, building functional abortions in these countries that can actually sequence bacterial genomes, because you need an extensive setup, you need a dedicated lab, tailored wet lab protocols, instrumentation, not to mention your sequences and your sequencing reagents. So that was obviously one aspect. Concomitant with lab capacity was addressing challenges with data, so data interpretation, you could easily generate data off the sequencer, but can you actually convert it to a form that is human readable and human interpretable? That we had to establish in terms of bioinformatics capacity at these various sites. And not just talent and capacity in terms of human resource, but also making sure that all of this data was actually processed in the cloud, because cloud-based pipelines means they lower costs, and anyone with a laptop and a modest wifi connection in these countries can replicate and run analysis for months on end. So that was really key to getting sustainability with bioinformatics.

And obviously the third aspect was, and this is something people tend to overlook, which is financial expertise. So most LMICs are typically underfunded, and unfortunately so, so we had to really work out a system where grant management and financial transactions are well managed so that they can be more sustainable with their operations over time. And we did this through a scheme called the Good Financial Grant Practice, which is a scheme that we developed with certifying bodies in Africa, and it was used essentially to improve financial mechanisms within these sites.
And I think the last block, and I guess probably the most important one was the actual training. So you can build as much lab and bioinformatics expertise as you like without any proper uptake, you don't have anything to deliver on. So we really devised a new method such as Train the Trainer, which is a program that we develop to provide specialists with not just the technical knowledge, but also the teaching skills to actually impart that knowledge onto their respective networks, because as founding members of the training network, we couldn't really go to all of the sites worldwide. That would just be impractical. So what we did was we not only gave them the technical expertise, but the way to teach others, which would be then exponentially propagated or self-propagated within their local networks.

Betsy Young:
That really makes a lot of sense. And public health laboratories, even here in the US I think could take a page from that guidebook. So thank you for setting that up. What would you say has been the most useful lesson from applying genomics to tackle AMR?

Mihir Kekre:
In my opinion, building an action plan against AMR, let alone a strategy that involves genomics can take over a decade. And we knew this going into it. It really takes even over 15 years really to pave forward at a population level. So there's obviously the investment in infrastructure, workforce training, building your resource networks, just to name a few upfront costs. So lessons wise, I think first lesson that we learned really early on was bring down the time needed to actually set up this infrastructure to manage infections. So get the systems up really, really quick in a matter of months to really create impact on tracking and surveillance. That was obviously lesson number one. And we captured this retrospectively in the form of an iterative approach that we recommend to assembling a surveillance laboratory. So that's something that came out as an output out of upfront.

The second lesson I think was making sure we share experiences. So we had to be super collaborative, share challenges, share our practicalities so that other surveillance networks can actually cookie cut from what we've done, but then also build and improve our models because we just did it as a pilot, so it may not be the best way of doing things and people will come up with better solutions. But I think the final most crucial lesson for me at least was how do we quickly show that the surveillance data you generate is tangibly valuable. People like people in government, regulatory bodies, and other healthcare providers are only interested in how tangible the data is, what utility it can provide.

So in order to sway clinicians and policy makers to use WGS in surveillance, we really needed to show that the data actually affects outbreaks, we're actually able to manage infection on a real world level. So that's definitely a lesson we've taken forward and I think it's something that we want to work on even more. We want to put LMICs, or low middle income countries, front and center of this for any future initiative as well.

Lydia Morrison:
I think you really pinpointed one of the crucial factors, which is the speed in which these systems are set up to allow those tangible results to be realized by the society and governments, so I think that was a really great answer. I'm curious what, if anything, you feel like we've learned from the COVID pandemic about the available tools and the capacity for large scale monitoring of disease?

Mihir Kekre:
As I mentioned earlier, contact tracing, disease monitoring, sanitizing, these have now become part of our everyday lives. In some respects, COVID-19, and this might sound a little controversial, has been a little less complex to address, in my opinion, because if you look at it's a single pathogen and it was represented as just an acute two to three year public health emergency. AMR, on the other hand, I think can learn a lot of lessons from COVID-19 for sure, but intrinsically it's a much more complex disease playing out over a very long timescale. So we live in a changed world now where large scale data and interpretation of disease patterns, of modeling outbreaks, predicting where the next infection's going to strike is almost as crucial as building your defense arsenal, for example, so it's biosecurity at its finest. So the good news is that I think the one thing I take from the COVID pandemic is its power to demonstrate the fact that swift international collaboration between governments, between R&D communities who are developing vaccines, for example, the swift collaboration has really been impactful in trying to get over the pandemic.

We sequenced millions and millions of coronavirus genomes across the UK only and interpreted how the strains are being transmitted, and this led to better control of the Delta and the Omicron variants, for example. Or just if you want to look at the speed with which we developed vaccines was incredible. We had the fastest approved vaccine for COVID-19 in about a few months, which was incredible. And obviously we had new ways of manufacturing vaccines, we had a lot of investment from countries, sure, but the real focus I think was all the regulatory bottlenecks and the political bureaucracy was addressed fairly quickly and resolved. And that's what I take from COVID that you can really apply to AMR. So governmental policy and public awareness really had a massive impact on fighting COVID pretty quickly, and I feel you can apply that to AMR as well.

Lydia Morrison:
That's a great response. And I think that you're right in pointing out that AMR is a much more complex problem because of all the strains that you're facing, and so I commend you on continuing to fight that battle and find that information, and help build the world's knowledge bank around how we can deal with preventing AMR and making the tools that we do have to fight bacteria and infection, making those tools last longer and serve us better, so thanks so much for your work.

Mihir Kekre:
Absolutely. It's been a pleasure working on the project, and hopefully we can continue on with more initiatives that address this on a global level.

Lydia Morrison:
Absolutely. Mihir, thanks so much for joining us today. It was a pleasure to have you.

Mihir Kekre:
Thank you very much for having me. It was an absolute pleasure to join you here today and just talk about AMR and obviously the work we've done over the past few years just impacting the surveillance community. Just to say, I think, and I hope you don't mind me plugging this in, all of our work from the last four to five years of the GHRU is published in a special supplement of the Clinical Infectious Diseases journal that I'd be happy to obviously share a link to. It just describes how we've actually implemented genomics as part of surveillance models at the various partner sites that I mentioned. But we actually go into detail on practical considerations, pro tips, pitfalls, roadmaps, protocols, that kind of stuff, and just really share our successes and challenges. And it's all open access, it's all free, and you can access all of the protocols if you ever have someone wanting to start up a genomics AMR laboratory. So I'd be happy to share the links so other people can build on the work we've done and maybe build even better models in the future.

Lydia Morrison:
That's great. Thanks so much for sharing that, and we'll be sure to include that link in the transcript of this podcast so all our listeners can find it there. Thanks again, Mihir. It's been really so interesting talking to you today, and please continue the hard work that you're doing. We all appreciate it.

Mihir Kekre:
Fantastic. Thanks a lot.

Lydia Morrison:
Thanks for joining us today, we won’t be releasing a new episode later this year, but we will be back in January with a fresh episode featuring Charlotte Houldcroft, who is a lecturer in the department of Genetics at the University of Cambridge and joins us to talk sequencing, from double-stranded DNA viruses to ancient DNA.


Loading Spinner
"