MAYA ADAM: 

Welcome to Health Compass. I'm your host, Maya Adam, director of Health Media Innovation at Stanford Medicine.

PURVESH KHATRI: 

We finally have the potential to bring precision medicine at point of care where you have the right diagnosis at right time, matched with the right treatment. Finally, that dream of precision medicine is becoming reality at point of care.

MAYA ADAM: 

Before he helped to develop a breakthrough test for one of the deadliest medical emergencies, Purvesh Khatri didn't have his future mapped out. In fact, he said that there was a time in his life when he really disliked biology and he wasn't at all sure where his career was headed. What he did have was curiosity and a willingness to try things that didn't always work out. That openness to pivot, to explore, to sit with uncertainty ended up shaping the scientist he is today. Purvesh now works on the acute medical emergency known as sepsis. It's a runaway immune response to infection that kills an estimated 11 million people worldwide each year, accounting for nearly one in five deaths globally. In the early minutes and hours of sepsis, clinicians often have to make life or death decisions without clear information. Traditional tests can take days. Sepsis doesn't wait. Purvesh and his team at Stanford Medicine have helped change that. They developed a new FDA-approved blood tests that delivers results in about 30 minutes, giving doctors clearer answers when time matters most. Today we're talking with Purvesh about the winding path that brought him here, the science behind his breakthrough test and why staying open to uncertainty may be one of the most powerful tools any of us have. Here's our discussion. Purvesh, thank you so much for making the time to join us today.

PURVESH KHATRI: 

Thank you for having me.

MAYA ADAM: 

I wonder if I can start with a bit of a personal story -- something that you remember from your life or your career path that sort of shaped where you are today.

PURVESH KHATRI: 

Yeah, that's interesting. I always say that I have this undiagnosed ADHD. I've never had had a straight line in a career. I started out as an electronics and communications engineer, and the biggest thing I remember about that decision is I did not want to be a doctor. There was a lot of incentive for me to be a doctor in India and I refused to be one, and I chose engineering. I changed careers in pre-Y2K bubble where everybody was moving towards trying to solve the Y2K problem and save the world. So I decided I'm going to become a software engineer. And I think that was one of the most interesting parts of my careers because that led me to where I am. Very quickly I realized I am not going to enjoy being a software engineer for the rest of my life. And the reason was it just sounded like I'll be working too much on finance. There was no social media those days. So I came to Stanford asking for very specifically something that was a mixture of software engineering and medicine. I wanted something. So in a way, I engineered that postdoc position where I asked for a 50/50 appointment in a wet lab and a dry lab, and that ultimately put me on a path where I am now. I got introduced to immunology. I don't think the way I think about immunology today would have been possible without me being an engineer first, where the electronics engineer in me, as I learned about immunology, started thinking about immune system as a distributed sensor with a built-in amplifier. So if you're trying to build any distributed sensor that senses the environment, what you want is exactly what the immune system has. It's distributed, goes to every organ. It has a built-in amplifier. When it sees that something wrong is going on, non-self antigen, it immediately amplifies itself, responds to the threat, and then it goes back to where it was, the homeostasis. And then it remembers it for the rest of your life. You have that memory, and that is, as an engineer, that's the best kind of sensor you can ever develop.

MAYA ADAM:

Interesting.

PURVESH KHATRI: 

I developed it, and that's how I started doing diagnostics and therapy that I do now.

MAYA ADAM:

Wow. So with that in mind, can you apply that -- let's jump right in here. Can you apply that to what's going on in the body during sepsis for our listeners? Can you explain that?

PURVESH KHATRI: 

So first, let's talk about what is sepsis. So sepsis, its definition has been evolving for the last 20, 30 years. The sepsis three definition now is dysregulated host response or the immune response to a pathogen. It doesn't say whether it's bacterial, viral, parasite, fungus. All it says is if the host is not responding the way it's supposed to, you have sepsis. Now, why has sepsis been a problem? So the reason for that is if you look at -- until now, we were trying to find the bug if you have sepsis. And we couldn't find the bug because what is one of the biggest function of immune response? Its biggest function, is to make sure that when a threat is detected, it does not go everywhere in the system that it is contained. So by definition, immune system is trying to make sure that the pathogen, which is in some organ, lung, heart, kidney, wherever, doesn't get into your bloodstream and go to other organs. And until now what we've been trying -- the field has been trying to do was look for that bug, and looking for a bug in blood is basically like looking for a needle in a haystack. The problem is there is no needle in the haystack because the haystack's function is to make sure that the needle gets into the haystack. And that was the real problem with sepsis until now. But how does immune system help us find this bug or this needle that is somewhere but not in the haystack? And that's where the evolution comes in. If you think about over the hundreds of millions of years of evolution, what was the function of immune system? Immune system's function is: understand what is self and not self, and whatever is not self, figure out what kind of not self it is. Is it bacteria, virus, parasite, fungus, what have you? And then over the hundreds of millions of years, it also learned how to respond to these different classes. Not specific bug, but a group of bug. And it has learned, and these learning, these lessons have been passed through evolution, survival of the fittest. Those who learn how to respond, they survive. So these lessons have been passed through generations and it's in all of us. So what we started by saying is, why are we going against nature to diagnose sepsis by looking for a bug when the nature has made sure that it doesn't become systematic? Why don't we ask nature itself, what are you responding to? And if we can do that, we don't need to worry about where the bug is, what the bug is. It basically tells you what is the class of a bug that it is responding to. And that's where immune system has learned, what is bacteria, what is virus.

MAYA ADAM:

And Purvesh, what does that look like in practice in the emergency department? How fast does it move? What are clinicians up against?

PURVESH KHATRI: 

So this is something that I did not know as an electronics and communications engineer turned software engineer. And what was really an eye-opening for me was every -- sepsis moves really fast. So this paper showed that every hour of delay in administering antibiotic increased the risk of mortality 6-8%. And that is a huge number where what this means is hours met hours and minutes matter, not days. This goes very fast. And that was an eye-opening thing for me. And when I asked why can we not diagnose sepsis fast, what I learned, we are still diagnosing sepsis by looking for the bug. That means you have to take the blood sample, send it out for a culture which takes anywhere from 48 to 72 hours. You can do a few other things, a few rapid tests, but aren't an accurate proxy for what's happening at an immune system level.

MAYA ADAM:

So explain to me again, maybe in a way that maybe a non-scientist might understand how the test works and what is the 29-gene signature and how is that applied to basically find that needle.

PURVESH KHATRI: 

Just as a background, what is this test? This test is called TriVerity. We received FDA clearance for it earlier this year in January 2025.

MAYA ADAM:

Congratulations.

PURVESH KHATRI: 

Thank you. And what this test does in a very simple way is it measures 29 genes in blood sample. Now, when I say the genes, it is measuring the genes from the host, the person who's infected, not the bug. It's not looking for a bug. It is reading 29 genes in peripheral blood of a patient where there is a suspicion that this person may have infection. There is no diagnosis yet, right? Somebody presented with an acute illness, fever, elevated heart rate. There is a suspicion that the person may have an infection. The nurse would take the blood sample, and you put the tube in a cartridge and put the cartridge in a machine and you walk away. So it's fully automated. It makes sure that the genes are, or the mRNA specifically, is extracted from the cell. It's amplified, quantified, runs the machine learning algorithm on it and it gives you three scores. Score number one is probability that you have a bacterial infection. Second one is probability that you have a viral infection because you could have only one or both infection. Many patients would have both bacterial and viral infection. And then the third score it gives you is the probability that the patient would need an ICU level care in the next seven days. So this is another distinguishing feature. Instead of simply saying whether this is a sepsis or not, we are also redefining the clinical problem on itself, where we are saying it's not enough to say this person has infection, but also whether can they go home or do we need to send them to a hospital and treat them faster? Why is this important? Imagine if we could have identified 80% of the COVID-19 patients during the pandemic and said, you can go home. We wouldn't have to have shut down the entire world and all the economies. We could have -- the reason we had to do that was law of large numbers. We were overwhelming the ICU beds, but we could have better managed with something like this. And that was a 30 minute test. It is a 30 minute test. No more waiting for the culture or what have you to come back. You can do this quickly, faster. And so I'm happy to give you some examples of how it's actually working out in the clinic with the combination of these three.

MAYA ADAM:

Please tell us.

PURVESH KHATRI: 

So this is one of my favorite examples that I just learned from one of the hospitals that has been using this test. And this was a 65-year-old came in with fever and then some complaints and doctors thought that this is infection, wanted to treat the person, but they're in TriVerity, and TriVerity result came back saying low viral infection probability, low bacterial infectionnprobability meaning the person is not infected, it's something else. And so they checked for heart attack and the patient was actually having a heart attack. It wasn't an infection. And imagine the life saved here in 30 minute and not give the wrong treatment and actually say, do this other diagnostic workup. It's not -- and then that's the other advantage of TriVerity. It's not only about infection, it also tells you, hey, this is not infection because I don't see bacterial or viral infection. It may be something else. And it allows you to trigger an entirely new diagnostic workup. And that's one of the things that's really exciting to me, that we can actually be at much earlier in differential diagnostic pipeline to figure out what is the right diagnostic, what is the right patient care? And this is all based on reading immune responses and not having to guess what.

MAYA ADAM:

And Purvesh, I'm curious if the immune system is sort of the common denominator here. Are there other diseases that this approach could help with?

PURVESH KHATRI: 

Oh, this is my favorite question. The answer is absolutely yes. We have applied this approach to organ transplant, pulmonary diseases like fibrosis, asthma, interstitial lung diseases. We've applied it to autoimmune diseases. We've used this for vaccine response predictions before somebody gets vaccine. And across the board, we have seen a few things. And the biggest and most important thing is we repeatedly see that immune system immune response actually turns on before a patient or a clinician suspects something is wrong. It is by design, by evolution, that immune system tells you early on that something is wrong. And I'll give you another example. For example, the sepsis and other critical illnesses, acute respiratory distress syndrome, burn trauma. And we just published a paper, it's the largest study of critical illnesses, anything that lands you in ICU, and we could show that all of the critical illnesses actually fall along a continuum of immune dysregulation. We define them based on syndromes, but we can show that underlying, it's the same immune axis that goes from being mild infection or a mild outcome to severe outcome, including fatality. And we could actually use that to say, because now we know which part of the immune system is not working, we can actually tell what is the drug that the patient would respond. And if I were to bring everything that I have said so far into a single summary statement: We finally have the potential to bring precision medicine at point of care where you have the right diagnosis at right time, matched with the right treatment. Finally, that dream of precision medicine is becoming reality at point of care. No more sending out samples. You can do it at point of care.

MAYA ADAM:

Wow. Purvesh, I wonder about adoption. Where is the test being used now and what are the biggest barriers you face to getting this test used everywhere?

PURVESH KHATRI: 

So about 10-ish hospitals have started using this so far. I'm excited to say the adoption has been way better than I as an academic expected given that we haven't even gone through one year since the FDA clearance. This is a first-in-class diagnostic, nothing like this has been in clinic. So one problem is education, actually explaining to people that yes, this is a sepsis diagnosis, but this is not like all the other ones. This is a different class. How does this affect the overall health economics? And we are working on that. But that have also been, there was one independent study out of -- it's a small study and it was retrospective done at UCSF, our colleagues up north. And they showed that on average, the test paid for itself by avoiding a number of other diagnostic tests that had to be run. So the average cost saving per patient for not doing other tests was basically the cost of the test. And in this, they did not include the cost of hospitalization, the ICU bed or a bed on the ward. All of those costs that are avoided, that was not included. We have another study, I won't say the name of the institute just yet, but we are writing up this study where the TriVerity was basically the intervention and we compared how the clinical practice and the outcomes changed before and after TriVerity was implemented. And what we found in emergency department is TriVerity reduce the ED wait time. So from the discharge -- usually people wait for hours -- it reduced on average by about five hours in ED. So that was number one. Number two, more people were sent home from ED instead of put on the ward for observation. So it reduced that expense. And then the question could be, well, if you send people home faster and more people, how do you know it is safe? So we followed them over, their outcomes, over the next three days, seven days, 30 days. And consistently we found that less patients, lower number of people came back to ED with the same complaint. So we were now sending more people home faster and safer. And what that means from the cost perspective is an overall reduction in healthcare utilization per patient.

MAYA ADAM:

So as I hear you explaining this, I see how animated you are and how excited you are about this work, but I was told that you didn't always love biology as a subject, for example, in school. So can you take us back for a moment and tell us how that shifted in your life?

PURVESH KHATRI: 

Yeah, so as I said, I had an opportunity to be a doctor or an engineer, and I still vividly remember telling my dad who was trying to convince me to be a doctor, and I said, come hell or high water, I am not going into biology. I still can picture myself exactly where I was when I told him that. So I became an engineer, I became a computer scientist, and I came to Stanford as a postdoc and Atul made it work for me so that I had a 50% appointment in a wet lab and it was an organ transplant lab. And I was trying to learn about what is an acute rejection or a rejection of a transplanted organ, kidney transplant. And I came across this really phenomenal paper called "The immunologic constant of rejection." Not organ rejection, just tissue rejection that includes autoimmune diseases or cancer, solid organ transplant and so on. So what it was talking about is that is this constant immune response -- or there is a conserved immune response that leads to tissue destruction, tissue rejection. And when I read that, it was very much the engineer and a computer scientist in me could understand that paper. One, it was very well written, but the second was I could see my training in engineering being very useful because what the paper basically said was yes, the triggers for immune response might be different because that's what immune system supposed to do. It senses your environment, not just the outside, but inside the body. And based on whatever it senses, it triggers the response. And what the paper said was yes, the triggers may be different, but ultimately tissue rejection, the tissue injury, is a common pathway. It's like all roads lead to Rome. Ultimately, everything goes through this one pathway. And the software engineer in me was like, yeah, that basically means if I'm a software engineer, yes, I might have a printout -- I want a printout of a webpage or a Word document or an Excel spreadsheet, but everything ultimately goes to the printer and the printer is what prints it out. Same thing is happening with the immune system. I may have solid organ rejection, bacterial infection, viral infection, cancer, but ultimately everything goes through this one pathway and we should be looking for that despite all the heterogeneity of the patient population. So anyway, so that's where I really suddenly I could see that biology is not how I was taught in India in a high school, which was all memorization. That immune system and nature evolution in general, there is a logic to it. There is a modularity to it, just like in a softwares that I was writing. And I can think about immune system as those module T-cell, B-cell, monocyte each one with specific function and they work together as a system that there are different function calls. Right? In software engineering there are these API calls, which basically means one part of the software calls another part of the software, and that's what is happening in immune system. Innate cells, monocytes and neutrophils and dendritic cells, they figure out what is the bug and they present it on the cell surface, and then the T-cells and the B-cell say, is this self, not self? It is not myself. So I'm going to say die. I'm going to kill that cell. It all sort of, now I think of biology as a software engineering,

MAYA ADAM:

It makes me wonder about the value of a non-linear path. I wonder if this work that you're doing and the discoveries you've made, if they would've happened, if you'd had a more traditional trajectory.

PURVESH KHATRI: 

For me personally, it would not have worked. I worked very much by making associations and it really helped me to have this very differentiated background in terms of what I do now. Because I still don't think about -- for example, in the early days of sepsis, there were a lot of people telling me sepsis is only bacterial infection. And the first time I said, "Hey, this viral infection looks like sepsis." I was actually told by a clinician that, "No, no, no, Purvesh, that is not sepsis, that's just severe viral infection." And sepsis has this. And I was given this- - And then at that time, this was 2014, so more than a decade ago, I sat back and said, huh, there is data clearly telling me this is what's going on. And somebody with years of experience is, I wouldn't say refusing to look at the data, but not able to see that the data is telling you something different. There are these preconceived subconscious biases. And that was one of the thing that I think has helped me, not being a biologist, I don't have preconceived notions. So many times when I present something or I write a paper, I get a response, "Oh, we've already known this for like 20 years." And I say, "That means I could recapitulate the last 20 years of work with a single completely unbiased analysis." And what that means is I am on the right path and I can continue working on that instead of saying, "Oh, this has already been done."

MAYA ADAM:

Interesting. Thank you so much for your time and for sharing your story so openly. Your journey is a powerful reminder that there's no single path to meaningful work.

PURVESH KHATRI: 

Thank you for having me again.

MAYA ADAM:

Thank you for listening to Stanford Medicine's Health Compass podcast. If you'd like to hear more conversations like this one, you can follow Health Compass on the Stanford Medicine YouTube channel, or any podcast platform that you use. Stay well and see you next time.