Safe and effective use of vancomycin

[Music] Welcome to the Australian Prescriber Podcast. An independent, no-nonsense podcast for busy health professionals.

Hi again, everyone. I'm Dhineli Perera, your host for this episode, and I can't tell you how rapt I am to be chatting to 2 pharmacists today who are considered experts in the world of antimicrobial stewardship and therapeutic drug monitoring. They were both on the Therapeutic Guidelines Antibiotic Expert Group. We've got Amy Legg and Tony Lai. Amy is a clinical pharmacist at the Royal Brisbane and Women's Hospital, as well as a PhD candidate through Menzies School of Health Research at Charles Darwin University. Tony is the Senior AMS [antimicrobial stewardship] pharmacist at the Children's Hospital at Westmead, Sydney, as well as a senior lecturer at The University of Sydney. Amy, Tony and their co-authors have written a practical piece demystifying the upheaval seen in vancomycin TDM [therapeutic drug monitoring] over the last decade and explained what this means for day-to-day application by our frontline clinicians. A really warm welcome to the program, Amy and Tony.

AL: Thanks, Dhineli. Pleasure to be here.

TL: Pleasure to be here. Thanks for the opportunity.

So Amy, maybe you can start us off with a little trip down memory lane. Vancomycin trough monitoring – how has that evolved over the last one to 2 decades?

AL: Great question, Dhineli, and it's such a pleasure to be talking about vancomycin. I think the story of the vancomycin trough concentrations is really why people get a bit nervous about using vancomycin because of these concerns about what the exact correct target is. So way back in 2010, the trough concentration recommended by Therapeutic Guidelines used to be 15 plus or minus 3. So going for trough concentrations around 12 to 18 [mg/L]. And over time, we realised that even though trough concentrations are easy to check because we just take a blood sample immediately before the vancomycin dose and then we just compare it to a set range that's all very straightforward, they're not actually predictive of the most effective way that vancomycin kills bacteria and vancomycin's pharmacodynamic parameter for efficacy is actually more about total exposure related to the area under the curve. So what we were trying to do was achieve an area under the vancomycin 24-hour concentration–time curve of greater than 400, and to assure that we were doing that with a trough concentration, we really had to start aiming for a trough concentration above 15.

And at that time, probably vancomycin was being used a lot less than it is now. With the rise of antimicrobial resistance and changes in global health, we probably are seeing a lot more MRSA [methicillin-resistant Staphylococcus aureus] in our antibiograms, meaning that we have to use vancomycin empirically more often. And so the increase in vancomycin usage coupled with the increase in target concentration of greater than 15 in the subsequent years led to this huge increase in acute kidney injury. And so now we're not happy to go back to less than 15 because it doesn't guarantee you the AUC you need and we're not happy to stick with over 15 because it has an increased risk of nephrotoxicity. So we're trying to be more direct about how we monitor vancomycin and not give any more than we need to.

Okay. So with the advent of AUC [area under the concentration–time curve] monitoring for vancomycin, Tony, could you help refresh listeners on what exactly an AUC is actually talking about?

TL: Yeah, sure. So the key word with AUC is exposure. So if you were to do a typical pharmacokinetic (PK) curve, which is X-axis (time), and Y-axis (concentration), and you drew a line of typical PK that goes up and down and you grabbed a marker and you shaded that area in, that's what we're referring to. And that's typically in most instances over a 24-hour period, but it can be over a dosing interval, but certainly within the chapters of the Therapeutic Guidelines, it's used over the 24-hour period and it's reported in these weird units like mg hours per litre (mg.h/L), but it integrates both the concentration and time perfectly over that period rather than a single time point, like a trough or a peak. And it correlates more closely with clinical outcomes better than those single time points like a trough.

Okay. And so then extending that, what does the ratio of an AUC/MIC [minimum inhibitory concentration] mean?

TL: An AUC/MIC is one of 3 types of what we call pharmacokinetic and pharmacodynamic indices, which can determine how bacteriostatic or bactericidal a certain antimicrobial may be. So there could be a peak over MIC. There could be a time above MIC and certainly for vancomycin, what most accurately exhibits its bactericidal effect is this relationship of AUC/MIC, the minimum inhibitory concentration. So it integrates that pharmacodynamic marker to predict treatment success.

Just understanding that ratio to begin with is kind of important right? Otherwise, they're just numbers we're throwing around.

TL: Yeah, exactly.

So then further again from this, Amy, what AUC/MIC ratio are we targeting for vancomycin?

AL: So the MIC component of the AUC/MIC ratio is tricky. The fundamental level, the MIC is telling us how easy or hard the bug is to kill that we've grown in the lab and the lower the number, the easier the bug is to kill. With vancomycin really initially based on MRSA bacteraemia, we've developed an AUC/MIC ratio target of 400, which is our floor, our target for efficacy with our ceiling being 600, which is related to an increased risk of nephrotoxicity. So really our target is an AUC/MIC ratio of 400 to 600. Now I'm going to start getting into the weeds a little bit with MIC now, so hopefully make our lives easier later.

MICs mean lots of things to lots of different people. In the lab, we have a breakpoint MIC that is trying to differentiate pathogens that have a resistance mechanism as opposed to the wild type bacteria that don't. And so those breakpoints have a specific meaning in the laboratory completely external to how we now want to use MICs for drug dosing. The trick about AUC/MIC ratios is they're only really related to efficacy. I know I just said that the AUC/MIC ratio for toxicity is 600, but MIC is irrelevant to toxicity discussions. So what we have to do is then try and work out what MIC we're happy with to be using these AUC/MICs ratios. Essentially, having said all that, we assume an MIC of 1 for vancomycin drug dosing, which means that our AUC target is 400 to 600. In the lab, the MIC can go up to 2. That's very infrequent. And in that situation, we would really worry about how we were going to achieve the vancomycin target we needed without causing nephrotoxicity.

Is there a situation where you'd recommend asking for the MIC to confirm that or can we always presume that it's 1?

AL: Yeah, I wish we could always presume it's 1. It's a real shame that the breakpoint MIC and the drug dosing MIC aren't sort of on the same plane here. Most of the labs will report the higher MIC. So for MRSA, most labs, once the MIC is above 1, it does get reported because there are data to suggest that drugs like vancomycin aren't as effective once the MIC increases above 1, and that might be because of exposure, it might be because of low-level resistance harbouring in the organisms. But I think in terms of pharmacists requesting an MIC, that's difficult. My experience traditionally has been the lab tells us that there's a higher MIC and then we make decisions based on that.

Excellent. And then the next question that was part of that for you, Amy, is there a higher target recommended for CNS [central nervous system] infections like we did for the troughs?

AL: Yes. So we've been largely silent in this edition of Therapeutic Guidelines on what the target exposure for vancomycin is for CNS infections. There's lots of diversity in what the penetration of vancomycin is into the CNS based on the patient, based on the permeability of the blood–brain barrier at the time. But I think overall what I would say is we don't have enough information to routinely use an AUC as part of our target for a CNS infection. In the first instance, there are some institutions that don't even feel comfortable using vancomycin anymore for CNS infections. And in some ways, giving a target means that the TG [Therapeutic Guidelines] were advocating that Vancomycin is the right drug for that, which wasn't the consensus view of the expert group.

My institution, lots of institutions do still use vancomycin for CNS infections. And what I think about is I know it has a reduced penetration into the CNS, albeit highly variable, but I wouldn't expect the same target for an MRSA bloodstream infection, when I have essentially 100% exposure, to a reduced exposure in the CNS. So I try and think about how can I safely give prolonged high concentrations of vancomycin? And often that means that we do aim for higher trough concentrations if we are giving it by an intermittent infusion, but still trying to minimise the total daily dose of vancomycin. So either going to 8-hourly and we pretty infrequently end up switching to a continuous infusion so that we can have the lowest total daily dose of vancomycin, but with prolonged high exposures to try and maximise penetration to the CNS.

Right. So without a target, would you suggest clinicians continue on as they had for CNS infections in terms of what they are aiming for and is that what you've done in practise?

AL: Yeah, great question. I think we've said something helpful in the guideline like ‘seek expert advice’. But yeah, essentially we used to always say to aim for a trough concentration of 20 to 25. Realistically, using 12-hourly dosing aiming for a trough of 20 to 25 gives you a huge exposure. So I think either using a continuous infusion aiming for 20 to 25 or intermittent infusion really probably at the higher end of the traditional trough concentration targets.

So Tony, we have these targets for efficacy, but what about toxicity? Amy has alluded to the fact that the AUC/MIC is almost used as a surrogate for that, but is there a single AUC that's likely to be problematic or sustained AUCs over prolonged durations? How do we work with monitoring for toxicity in this new space?

TL: The AUC in relation to toxicity and the threshold is a spectrum of risk. You'll see that nice graph that we've put into the Aus Prescriber article, which shows it quite well. There's always going to be a risk of nephrotoxicity just by giving vanc and having low AUC. So it's just more exposure will increase your risk, but it increases your risk significantly when it goes over generally the 600 mark AUC in a 24-hour period. Now, in terms of whether or not it's a single high AUC over 24 hours or sustained, it's generally considered to be more of sustained effect. But I need to caveat that in saying that this is based on classic definitions of nephrotoxicity, which is used throughout the literature, but it is actually quite crude. So they use baseline creatinine and whether or not the creatinine increases by 1.5 times that. Now creatinine itself is probably a delayed marker of nephrotoxicity.

So there are markers like Cystatin C, which shows renal stress, but by that definition, and if you look at the papers, the creatinine tends to separate. There's this really good paper by Finch et al. where the group compared AUC versus troughs and they had a COX proportional hazards graph. And at day zero, both lines overlap, but then when it reaches about day 2, both lines begin to separate where troughs seem to have higher association of nephrotoxicity and AUC kind of stays above it. So the inflexion point occurs at day one to 2.

But as I said, that might be because of the definitions we're using of nephrotoxicity. It might actually be earlier, but if we're using creatinine, that's what we use at the bedside. This is what clinicians use day to day. It's generally around 2 days, which is actually a good thing for clinicians and pharmacists in this space because we can action the TDM during that time. So you don't really need to be too urgent in measuring AUCs pumping into a Bayesian forecasting software on day one. It could be day one, 2, and generally you need to wait for the steady state level, which is after 24 hours anyway. So it's certainly more sustained rather than a single and more prolonged is what increases your risk of toxicity.

Okay, that makes sense. So continuing on with you, Tony, when it comes to monitoring and dosage adjustment, it sounds like there are some cases where a hybrid of trough monitoring and AUC monitoring are needed, like in rapidly changing renal function or as we mentioned, CNS infection. Is this correct? And are there others where you would use a hybrid?

TL: Oh, certainly, yeah. When you look back at the PK model working in the background of the program you're using, if that patient population was not large enough and included the type of patient that you are treating on the bedside, that's probably when you should consider more of a hybrid approach because we often treat quite atypical patients, may it be patients on ECMO [extracorporeal membrane oxygenation], obese, renal replacement therapy, and sometimes the PK model won't describe their pharmacokinetics as well as if you were to do it via a hybrid method where you would be using more concentrations and associating that with the dynamics of the patient.

Yeah, it's not a case of plug the numbers in, spit the result out and apply it. It really is looking at the model fit and how closely the model population on your program is reflecting the patient that you've got in front of you.

TL: Absolutely. Yep.

Continuing on with that AUC path, Amy, we use an adjustment table I know in our institution for ‘the trough is this so adjust it by this amount’. Can AUC monitoring be used to guide adjustments as well? So perhaps there's an AUC of 650, this means we need to reduce the total daily dose by x percent. Does that exist?

AL: Yeah, it does depend a little bit how you're calculating your AUCs because I suspect the vast majority of pharmacists in Australia are going to be using a dosing software, in which case essentially the dosing software will work out with you what an appropriate dose would be. So a lot of the freely available, but especially the subscription ones, you can tell it what you're aiming for and it will put together some doses for you to check. Vanc is quite linear if the renal function stays the same. So you could come up with your own dose based on a linear adjustment of the AUC, but realistically, the beauty of the AUC-based monitoring is actually that part of the work. The dose adjustments is much easier because the very complicated computer system is doing it in the background.

Yeah, excellent. And so Amy, with the methods of AUC monitoring your paper had a really neat table describing the advantages and disadvantages of the different methods. Could you cherry-pick the pertinent difference for us?

AL: Yeah, I do get asked this a lot as well. So people are trying to work out if they're going to switch to AUC-based monitoring, what's the best way of doing it? So there are some pharmacokinetic equations. They're essentially just less sophisticated than the Bayesian dose optimisation software. The pharmacokinetic equations do normally require 2 levels, and then they just use a fairly basic trapezoidal method to work out what the AUC is. A nice example is the Sanford calculator. It needs to be at steady state because essentially you can work out an AUC for that day, but the program doesn't know whether it's going up or down. The program's just working out the AUC for that day based on those concentrations that you've put in. So they're very easy. They can still provide dosing recommendations, but it normally takes 2 levels and you need to be at steady state or you need to be able to interpret what you are telling it based on the limitations of the calculator.

The Bayesian dose optimisation software have more detailed data entry requirements and then use essentially more complicated mathematical modelling in the background to work out what the concentrations were, where the concentrations are going. You can visualise a concentration–time profile, you can simulate new doses, you can simulate what the trough is going to be with a certain AUC, and they really help educate, even the clinician on vancomycin as you use them. They're not all free. They don't all have TGA [Therapeutic Goods Administration] registration, whether they need to have TGA registration is a bit of a matter of debate at the moment, but they are certainly providing a lot more information. And this idea about are you at steady state, the Bayesian dosing software will tell you exactly where the patient's going. The one sort of fiddly thing that I have been playing a lot with in my mind is whether a lot of sites are going to end up probably with a hybrid of AUCs for high-risk patients, trough concentrations for lower-risk patients where risk is both risk of failure and risk of toxicity.

And to do that, it's nice to keep doing troughs for everyone. And then for some people using this lower trough target of 10 to 15 if they're low risk, and then for other people plugging whatever the trough is into a Bayesian or a pharmacokinetic equation and working at the AUC. Troughs may not be the best time point for calculation of an AUC because it's just anchoring the concentration–time profile to one point at the end of the dosing interval.

Having said all of that for now, we've said that you can use a trough concentration into the Bayesian dose optimisation software, but maybe over time we will end up with a different time point as our optimal time point for that calculation.

That's really interesting. Thank you for sharing that. So Tony, who should AUC monitoring be prioritised for?

TL: The number one answer I would say is patients with MRSA bacteraemia. They're the type of patients where AUC is most critical because you know that they're going to score a decent duration of vancomycin and you know that MRSA bacteraemia is associated with high mortality, 10 to 20%, but also patients who will be on vancomycin for long durations and they might not even have a bacteraemia, so it could be osteomyelitis, for example. Certainly endocarditis fits into the bacteraemia space, and sometimes we would give vancomycin for long durations for a coagulase-negative Staph, which is like the weaker brother of MRSA.

So again, because we know that the prolonged AUCs above 600, that's going to flick you over to the higher risk of nephrotoxicity. They're the ones you need to be a bit more comprehensive in terms of the level of TDM you do. So they're the ones whereby you should do AUC monitoring. And certainly in my organisation, when patients who are on prolonged empirical therapy, so febrile neutropenia, patients with lines in ICU [intensive care unit] who are generally on vanc a bit longer, I would probably do an AUC in those type of patients where I see they've done one or 2 dose adjustments based on trough-based monitoring and they're not hitting the target, we'd probably achieve the target earlier, save this patient from doing extra bloods and save clinician time. So those are the types of patients that would be number one in my mind.

Yeah, for sure. And then bringing your paediatric expertise into it, I'm going to assume you're going to say a hard no to this in terms of can we directly transfer our AUC adult targets to our paeds patients?

TL: So vancomycin PK is very different in children. In general, kids above 3 months to 16 years, they clear vanc faster, generally need high doses if you were to look at mg per kg per day, and in the paeds space, we give more frequent dosing, so generally 6 hours versus adults every 12 hours in a patient who has normal renal function. And because of those different layers of how we give vanc, the doses required, the targets are reflective of that, and they might be slightly different. So because we're giving more frequent dosing to get the same total 24-hour AUC, the troughs are kind of on the even lower side from 7 all the way to 15, so what happens to paeds? Paeds treat vanc differently and we should treat them differently accordingly.

And then finally, Amy, it would be remiss of me not to ask if there are any other major changes in the updated TG antibiotic guidelines that you'd like to highlight here for our listeners.

AL: Yeah, thank you. So a couple of the other big changes, one is who needs a loading dose? Really loading doses help us to get to a therapeutic concentration within the first 12 hours, and that's particularly important for patients who are critically unwell, who have a high risk of mortality, and you don't want to wait 36, 48 hours to get therapeutic. So we've largely removed the recommendation for a loading dose outside of critically ill patients.

The other thing that we have introduced in this edition of the guideline is Vancomycin recommendations for dosing in intermittent haemodialysis. The other group that often comes up is, what about dosing in obesity? We still recommend mg per kg doses based on actual body weight. We've introduced dose caps, and so for a loading dose for our critically ill patients, we would cap the dose at 3 grams, maybe slightly lower if they've got significant kidney impairment. And for intermittent doses, we're really talking at capping the dose at 2 grams.

Lovely. Well, that's unfortunately all the time we've got for this episode. Many thanks for joining us today, Amy and Tony.

AL: Thanks so much.

TL: A pleasure.

Amy and Tony's article, ‘Safe and Effective Use of Vancomycin’, is available on the Australian Prescriber website. The views of the hosts and guests on the podcast are their own and may not represent Australian Prescriber or Therapeutic Guidelines. Amy Legg has no conflicts of interest to declare. Tony Lai has accepted a Pfizer research grant unrelated to vancomycin. I'm Dhineli Perera and thanks for joining us on the Australian Prescriber Podcast.