Latest / CRITKNOW / Beyond the Data: FAST

Transcript

Beyond the Data: FAST · CRITKNOW

• Bharath Kumar T V: Hello everyone and welcome to the second episode of Beyond the Data. This is a follow up to our session on the FAST trial. We're delighted to have with us as our guest Dr. Nandini Sethuraman, who's the senior consultant and chief of microbiology at the Central Reference Laboratory of Apollo Hospitals in Chennai. She is also the site investigator for the ICMR regional AMR network and has a particular interest in antimicrobial susceptibility testing. So it's really wonderful to have you here Nandini and you know we're looking forward to hearing your thoughts not just on the trial but everything really around it. So thank you for making time. Nandini: Likewise, I'm really happy to be here with you both. Ramesh Venkataraman: Yeah, thank you, Nandani. Again, I echo everything what Bharat said. So I I think ⁓ you would bring a lot of clarity to a lot of these techniques that we commonly talk about. And I think ⁓ it'll be interesting to hear your perspectives on how you see the study and where we should go from here. So with that, Nandani, my first question to you is ⁓ could you just go over the steps of how if a blood culture is sent to you, how do you First of all, identify positivity of blood culture using the traditional methods. And are there newer methods that expediate this process of basically answering the first question? Is there a bacterial growth or not? So could you just tell us how you go about doing it, Nandi? Nandini: So the traditional workflow of blood cultures. once they are collected into the bottles. So I'm going to stick with automated blood culture systems because predominantly that is what we have available in most of the centers which have ICUs, et cetera. So the bottles get loaded into the machine. Now, right there, there is a time gap. For example, if it's a lab that is running 24-7, the bottles are loaded into the machine throughout. But there may be cases where the labs are shut in the night, so the bottles need to be kept out. So there can be about an 8 to 10 hour delay. over there. Assuming it has been loaded into the machine, usually it takes anywhere between 6 to 9 to 6 hours, the 5 day time period for which we keep the blood culture bottles inside. But if there is a positive and if there is a gram negative specifically, it usually flags within 24 hours as early as 6 hours is also possible. By flag, I mean the instrument sort of gives a beep and the bottle comes out of the machine or you can like take it out particularly that bottle out of the machine. And as soon as we take it out, the first thing that we do is a gram stain from that blood culture broth. We take a little bit of that broth from the bottle, put it on a slide, do the gram stain, and see what is the bug, whether it's a gram positive coca or a gram positive bacilli. Simultaneously, we will also do a subculture onto solid media, which will then take overnight to show up colonies. So we do an overnight incubation from those plates on which we have done the subculture. So that's again an overnight 16 to 18 hour gap that happens over there. Then from the colonies the next day, this is all I'm about the traditional algorithm. From the colonies, we would set up usually simultaneously an identification and a susceptibility testing. Ramesh Venkataraman: So just to pause you there, Nandini, when you say it flags, ⁓ my understanding is there is some CO2 emission that happens and that is what is being detected by the sensors. Is it the same thing for whatever bacteria that grows or is the sensing like CO2 molecule? ⁓ rather than CO2, it senses something else for gram positive, something for gram negative, or is it just it's it's maybe a dumb question, but I'm just asking so that we are very clear how it flags. Nandini: Yes, yes. Yeah. It yeah it is the same for all living things almost. In fact, I say living things because if there are if there is a very highly neutrophilic sample it will flag false positive sometimes. So bacteria, whether they're gram positive, gram negative, fungi, yeast, everything will grow in the same system with the same mechanism. So the same model is good for all of these organisms, usually, and even the more fastidious ones. The only thing that changes is the time to positivity. Gram negatives generally are higher in opculem, faster growing. So they will be faster than, say, for gram positives. And gram positives are faster than yeast. Yeast will take about 48 hours to flag, usually. And very fastidious organisms can actually take to four or days which is why we keep the bottles in the system for five days before we call it negative. So that's the mechanism. Bharath Kumar T V: Right, right, right, Ramesh Venkataraman: Are there any new techniques that expect the identifying of the bug? Like you said, you had to go through the subculture and all that is that are there any new techniques to only identify the bug faster than Nandini: Yeah. late. So maybe let me just complete the traditional algorithm. So once we get the overnight colonies, we set up an identification and susceptibility testing. That identification ⁓ in the best of labs today happens with MALDiTOF, which takes about a few minutes. But traditionally, you would set up a battery of biochemical tests, each of which will be positive and negative. Or to automate that process, you do something called a biotech or a Phoenix and automated biochemical Ramesh Venkataraman: Yeah, sure, sure. Bharath Kumar T V: Yeah. Nandini: based identification system which can take again at least 8 to 10 hours for the identification to come. So, now at various steps there are now techniques available that can shorten that turnaround time. ⁓ Of course, the one that we would really love to have is something that you just tell us directly from the blood sample you know what is there in it. Unfortunately, we don't have something like that which is well standardized as on today. Most of the techniques that we have today first depend on the blood culture flagging positive which can which anyway will take a minimum of 6 to 24 hours. what we can do is a short MALDI like we can use the same MALDI TOF technology that we use from colonies but we don't have to wait for it for the colonies to appear rather from the broad directly you can sort of process it do a differential centrifugation or a filtration step or just put it on a chocolate agar couple of hours incubated and it will show us the scum of growth and from that you can do an identification so that shortens identification by about 16 to 18 hours usually and then you have these Bharath Kumar T V: You can pull it. Nandini: gene based PCR based techniques for example, the BCID which is commonly available in India. It has a bunch of target organisms which will be identified on the ⁓ from the positively flagged blood culture. So, as far as identification goes these are the things that you can do faster and reduce the turnaround time. Bharath Kumar T V: Right, right, right. I think that's very, very informative. I mean, at least most of us as clinicians have forgotten what happens once the sample reaches you. We think there is some potentially magic. And we obviously want it as early as we can. But there's obviously all of these processes that need to go. and you highlighted how some of these steps are at least being sort of expedited by some of these newer technologies. I think that nicely sort of brings us to the next question. Now you have identified the organism, but obviously we also want to know the susceptibility. So maybe you can again walk us through what is the traditional methods of testing for susceptibility. ⁓ That is the first part. And then the second part is you know, considering this particular trial that we're going to talk about, what is this concept of a phenotype-based testing or a genotype-based testing? So maybe both of those questions. Nandini: Right. So I'll take you back to the traditional AST. Once you have those colonies that have appeared, once you've done the identification, then you set up a battery of antibiotics for susceptibility testing. So in the standard AST workflow, what we do is either a disk diffusion test. or a broad micro dilution test or automated susceptibilities which something like Bytex and Phoenix you have some automated platforms where you can do susceptibility testing. But the first the starting point of that is the overnight growth usually. And these standard techniques are all phenotypic in nature. By phenotypic it means that you are dependent on whether the anti whether the organism grows or not in the presence of an antibiotic at various concentrations. I mean we are getting a bit technical there. So, depending on the MIC of the organism at a particular concentration it will not grow and that is what we call it as the MIC of that organism and once you have an MIC. Ramesh Venkataraman: Antibiotic. Nandini: you interpret that against guidelines like CLSI or UCAS and that is a traditional which is typically phenotype based meaning that you need the organism to grow it is not reliant on the mechanism of resistance. ⁓ When we want to look at the mechanism of resistant, you generally need a genotypic test. So, the genotypic tests are the common ones that we use again at the BCID which looks at some of the resistance mechanism determinants like an ESBL or a carbapenemase like your NDM, KPC, WIM, ⁓ OXA48 things like that. Now, again these tests the genotypic tests are faster. and you can perform them on a positively flagged blood culture bottle itself without having to wait for the colonies to appear. So phenotypic tests will take time because they are growth dependent but they are independent of the mechanism of resistance. Bharath Kumar T V: Right. Right. And when you said in the genotypic that you're identifying the resistance mechanism, ⁓ that is, mean, what is sort of the, I guess, the false positivity or, what, like, do you, if the resistance is there, is it going to be like, you definitely know that that gene is being expressed, so you don't, should move to a, you know, a different antibiotic? Nandini: Yeah. So, that is the reason I said phenotypic is irrespective of the resistance determinant. So, occasionally you do see the genes being positive on the genotypic test, but the bacteria is for some reason not expressing it. So, you find an ESBL these things are rare of course, with especially the common genes that are present on the targets that we commonly use, quite rarely we do see something not expressed. Bharath Kumar T V: Right. Mm. Nandini: ⁓ On the other hand, you also have, you have only a limited number of targets on the BCID, right? So you may find a Klebsiella, for example, which doesn't have any resistance determinants that you find, but you actually see that it is an ESBL producer because there is only a CTXM on the panel, but there may be other mechanisms like SHV and TEM, which could have actually led to that ESBL. So most of the time when you have a shocky patient and you have no resistance genes, you still may not be able to rule Bharath Kumar T V: Right. Nandini: out resistance, but you can definitely rule in and escalate based on these genotypic tests. Ultimately you have to rely on the phenotype and whether the resistance determinant is actually expressed or not. Ramesh Venkataraman: So you do both parallelly, one to flag it early and then one to cross confirm just in case the genes don't light up but the resistance still exists later, right? You you would normally do both parallelly, phenotypical and genotypical. Genotypical would come back first, phenotypical come back later to confirm whatever you did. Is that Nandini: the So if you're asking me what I do, we kind of do that as a reflex test. But most labs would generally just do the standard identification and susceptibility testing. Where the labs have incorporated these rapid tests into their workflows, some of these things get shortened. Ramesh Venkataraman: So when you say shorten Nandini, what is the average time differential that you would generally see between a standard flow of the entire like the SOP that you clearly said from the time the blood is sent to you till you identify the antibiotic susceptibility versus a fast technique? How much time could you think we could save as clinicians? Nandini: Yeah, so traditionally it would take about 48 hours if you are being optimistic, like for the overnight colonies to occur and then the susceptibility. So 48 hours. With the faster techniques, faster genotypic techniques, it is reduced by at least around 40 hours. Faster phenotypic techniques, I think which we didn't really talk about earlier. ⁓ The faster phenotype, there is a rapid AST technique that you can directly do from the positive blood culture bottle. And then there are platforms like the YTech Reveal, which was used in this trial, and platforms like Accelerate Pheno, which also can be done on the Positively flag. Those are, again, phenotypic. Those will reduce to a little less, maybe about 30 hours from the traditional. So 30 to 40 hours of gain you can have depending on the kind of rapid technique that we use on the Positively flag. Bharath Kumar T V: Right. Ramesh Venkataraman: Do you want to talk about the rapid technique used in this study? Because you mentioned it ⁓ Nandini saying that ⁓ you know the the fast study used the white acryl technique. So what is it, how is it done? What are these rapid phenotypic techniques? Nandini: Yeah, so Vitek Reveal is yet to be launched in India. I think the Indian site in the trial had it placed just for the purpose of the trial. So Vitek Reveal is a test that can be done on the positively flagged ⁓ only if it's a gram negative. So there are certain caveats there. The way it works is it is phenotypic in which it basically detects those volatile organic compounds that are released normally in the metabolism of bacteria. And in the presence of antibiotic concentrations if that VOC production is not there you know that the bacteria has been inhibited. So, the system basically detects those those senses those volatile organic compounds to tell us whether this organism is inhibited at a particular concentration or not. Vitek Reveal ⁓ basically has 10 organisms on the panel. There are 10 species on the panel ⁓ which has a bunch of entrobacterials including E. coli, E. coli eclipsiella and some entrobacteria, etc. It has acinetobacteria, it has sodium monaziridinosa. So there you basically have to narrow it down to only those. So normally if you have to use Vitek Reveal in your ⁓ workflow, once the blood culture is positive, you do a gram stain and you see that it is negative, gram positive has no point of doing a whitetec reveal because it is not on the panel. Gram negative and initially at the gram stain itself if you are able to tell that there is only one kind of gram negative you can proceed with it which is usually not practically possible. You may see you know all of them look the same, but tomorrow the next day you will find that there are two or three different polymicrobial infections happening there. And once you set up the Y-TEC reveal you also have to simultaneously do a quick identification because the Y-TEC reveal can give you answers in 6 to 8 hours, but the interpretation of the Y-TEC reveal test will still require an identification. that identification then becomes the rate limiting step. To me it's like you cannot use white tech reveal in the absence of Malditov and also a rapid Malditov So those two really have to go hand-in-hand if you have to use the white tech reveal system. It's a good system But yeah, we are yet to have practical hands-on experience with the system, but these are the things that I foresee will be ⁓ Once we start Bharath Kumar T V: Great, great. technology right right right. a challenge, right? And I mean, I think obviously there is a cost differential to all of these rapid techniques, ⁓ at least in our context, and this is going to vary between sites within countries. But for us, what is the cost differential of sort of adding on, you know, in rough terms of adding on all of these, not by tech review, I know you don't have it, but other things like MALDI or even the regular white tech KST that you do, what sort of is the cost differential to any of these techniques? Nandini: Yeah. So if I have to go technique wise, think Maldi once you've just bought the equipment, the running costs are very cheap. The running costs of doing an identification with Maldi is probably cheaper than doing your conventional battery of biochemicals or doing a white. But these, but in general, the faster techniques come at a cost. So if I have to just compare and block the standard identification AST versus a rapid technique. Bharath Kumar T V: Right, right. Nandini: either genotypic or phenotypic, it will be 2 to 3 times more. per per test, but that is just cost per test. the ways in which you, I mean, it depends on how you're looking at cost. So ⁓ depending on how you incorporate it into workflows, are we doing rapid tests on all the blood cultures that flag positive? So you're probably wasting a lot of, know, for gram positives, it may not be necessary most of the time. For community acquired the bacteria like typhoid or falmanilla, you probably don't need to do a, you know, such a rapid technique. probably don't need it for Bharath Kumar T V: Yeah, yeah, yeah, absolutely. Hmm. Nandini: You don't need it for all patients. can if you are able to just selectively use it for critically ill patients and sepsis where it actually makes a difference I think you can come down on overall cost but also the opportunity costs and indirect costs of how much you save on unnecessary antibiotic usage and number of days in the ICU that and all if you start quantifying there are studies that have done it it actually works out equal or cheaper Bharath Kumar T V: Yeah, yeah, yeah. Antibiotics, yeah. Correct. Right. Yeah. Yeah. I mean, obviously cost is complex. It's not just the what either the user pays or what the yeah, absolutely. Yeah. Nandini: Yeah. Ramesh Venkataraman: We are only seeing the up upstream cost, right? We we it's very difficult to capture the downstream cost of our Nandini: If there are studies that have tried it and actually showed that you know you are not wasting more money at the very minimum it's not more expensive than not doing the test. Bharath Kumar T V: Right. Ramesh Venkataraman: It's wonderful Nandini. I think it was a very crisp overview of all the techniques that we commonly use and I think it would help all our audience in trying to understand what goes behind the blood culture processing and everything. Now quickly coming back to fast study, Nandini. I think we talked about the White Ech reveal and how it's being used in this study. What what thoughts do you have about the paper and as a microbiologist and where do you think this paper fits in our practice? Nandini: Yeah, so I actually requested Bharat if I can sit in on the podcast, but then, you know, he didn't want to bias me. But I think, yeah, I mean, as a trial, it adds to the evidence that rapid tests do make a difference, at least if not, you know, mortality outcomes, at least in some of the process measures like another time to appropriate therapy, and especially in the more resistant subset of infections. But I would have been actually happier to see that it actually made those differences. And I feel that ⁓ The outcomes, as we expect, did not happen because of the pragmatic nature of the trial. I think they tried to bring in a lot of real world variables. And because even between the sites, for example, if you look at the supplementary material, there was a lot of difference in how they process, how they enroll the patients, first of all. Some sites only enroll patients for which the blood cultures flagged in the morning shift. And some of them were working 24-7. The testing was done in real time but the information was only going on the next working day or the next Monday if it was a weekend so things like that. I think all of that would have introduced a lot of bias into the way into probably the ultimate outcome measures and if you look at the sites also there's a lot of variability in their local epidemiology. So there is a disproportionately high number of patients enrolled in Greece and in Israel. So Greece has a very high carbapenem resistance rate of similar to what we see in India. Israel has almost none, it is like zero. And India having such a high carbapenem residence rate only enrolled about 93 patients. So I think there is also a very great difference in the local epidemiology of the patients that were enrolled. And we spoke about the technology itself. I mean, you saying Vitek reveal and extrapolating that to all rapid techniques may not be also appropriate because it has limitations of its own. So even in the study, think 20 % of the gram negat- bacteria, they had two sort of, they could not use for final listing because they were in panel, ⁓ microbial, ⁓ things like that. and I had some observations about the drugs that were given in the initial three days ⁓ in the patients. And if you see that there are not a lot of CRE cover drugs there, which should have actually worked in both in Greece and India. Greece also has about 30 percent, I mean, if you look at other papers that have done that Ramesh Venkataraman: Of panel, all organisms are twenty percent, yeah. Bharath Kumar T V: Hmm. Nandini: It has about 30 % metal abetalactamase, many of it is NDM, even though KPC is their predominant carbapenemase. When you have such a lot of NDM, and I didn't see any of those NDM, my MBL covering agents like, know, ceftazity, maybe Bactam plus Astronome. This was probably before the Astronome, maybe Bactam era. But you don't see a lot of that. I don't know how effective the stewardship would have been. if I want to know what difference this test makes in a given patient, I would have happier if it was a much more controlled trial you know I mean you guys are the experts but you know the actual benefit of the test I think we have not been able to capture with this design of this pragmatic nature of the trial but this is of course the first time that they have done this kind of a trial in areas where there is high resistance so ⁓ that's a definite step in the right direction but yeah with all these things would probably have led to poor outcomes but then I don't think we should discount the value Bharath Kumar T V: Yeah. Nandini: of these tests just based on the results of this trial and definitely better design studies in the future can actually bring out those outcomes better. Ramesh Venkataraman: Well that's very well said, Nandani. I think ⁓ I think the key take-home points from me, from whatever you said, is this is only one technique. Don't club all rapid antibiotic susceptibility testing and then just ⁓ make your judgment based on this paper. So for this technique, like you said, there are a lot of logistical limitations that could have skewed the outcomes. We have also talked about it in our podcast earlier in terms of the chain of events that have to happen or align. For you to have a good outcome after you have given your results, right? I mean, you can only give the results of the antibiotic susceptibility, and then the clinicians is clinicians and the system have to act appropriately to get the patient better. So there are so many other factors that could have dinged the study. But like you said, no kudos to the authors because ⁓ like Bharat also said in the podcast, the first time that they have actually taken a test and actually evaluated clinical outcomes in a very relevant population. And I've done it very pragmatically so that I can think of it and say, okay, this is how I would have done it in real life. These are all the pluses and minus. Maybe they were underpowered ⁓ despite their ⁓ no ⁓ whatever calculations they did to show the outcomes for various reasons. So I I I think those are the take-home messages. And the second point you made very valid is don't use these tests indiscriminately, right? You need to be able to use it in specific situations to get specific information that you think would help the patient rather than saying, I'm going to use it in everybody because I have the technology in my in my hospital or my institution. I think both of those are valid points. Don't overdo at the same time, don't throw them out saying these are useless. I think the balance seems to be very important. And I I mean Working closely with microbiologists and infectious diseases actually helps us in our ICU. I mean, we work with you all the time and definitely that information helps you to take back and any early information is better for us to be able to fine-tune the antibiotics and to show outcome differences going to be hard no matter how big a trial you do. I think it's going to be very difficult because the system is chaotic beyond that. So it's going to be very difficult to find a signal in chaos. So that's I think my take on whatever you've said. Bharat, you have any thoughts? Bharath Kumar T V: No, think, no, nothing brought out a lot of points that that are additional to, you know, some of the things that we have spoken about on on our on the episode. ⁓ But I want to maybe ask from my perspective, one final question. So, I mean, if there is benefit of such a test, it appears to be for the cephalosporin resistant group or the carbapenemesis and try to get the carbapenem resistant group. Ramesh Venkataraman: Yeah. Bharath Kumar T V: So if you had Bytec Reveal or some other technology like that in your unit, how would you use it in terms of your workflow? Because it seems to be helping only in those cases. Of course, we have a huge problem of carbon resistance. But if you have to integrate such a technology, how would you do it? And would it further truncate these timelines to maybe under 48 hours? How do you see it evolving, I guess? Nandini: Yeah, so I think, mean, in the future, if it does become available, ⁓ like how we have done with the reflex testing of blood cultures with BCID, we have basically come up with an algorithm of where we can choose to do the test because we have sort of bundled it together. don't put this, we take the responsibility to decide which patients get a BCID based on that. Bharath Kumar T V: Right. Nandini: So that's basically a bunch of things that we do. We call it test diagnostic stewardship, just like it's the lab side of the antibiotic stewardship. So I think it is most useful in patients who are critically ill having sepsis where the timing of the test will actually matter and make a difference. ⁓ Probably not so much for community acquired infections right at the beginning. ⁓ I would probably use it in those those patients and also where the result will actually make a difference. we want ⁓ you can't like over diagnose your way through a deficient pharmacy. So you if you only when you have the kind of drugs available with you again in this fast trial the one of the things that they didn't have was salbactam, derlobactam in India and Greece for example. They saw a lot of acinetobacter which they couldn't use appropriate drugs for. So I think doing the test really matters in areas where you can actually act on those results in patients who will actually benefit from the test. Ramesh Venkataraman: Wonderful Nandani. I think you're ⁓ having you has been a pleasure. You know, very clearly, crisply, lucidly, succinctly summarize the entire workflow process for a lot of our audience who may not be familiar with the various techniques you you use. Now we kind of can put the whole study into a template of where the study could be applied and in terms of workflow, like you clearly said and your perspectives have have. have really been no different than what we would think from the clinical side. So I think ⁓ no thank you so much for spending time with us. No and we enjoyed interacting with you and learned a lot of new stuff from you. So thanks a lot, Nandine. And with that ⁓ we'll the microbiologists have the final word and and say no thank you again and ⁓ no see you all in the next podcast. Bharath Kumar T V: Yeah, absolutely. Nandini: Thank you. Bharath Kumar T V: Thank you.