Hello everyone, my name is Kirsten Kota and I'm a medical and surgical trauma ICU pharmacist at the Mayo Clinic in Rochester, Minnesota. I also have a chunk of my time that is set aside for ICU, emergency department, and urgent care antimicrobial stewardship. Today we'll be focusing on two main concepts within optimizing antimicrobial drug use and critical illness. First, we'll understand what baseline physiology and antimicrobial interactions look like and then we'll interpret what to do with antibiotics in the context of the acute physiologic changes of critical illness. Secondarily, we'll take an assessment of that stage we're really more, you know, two, three days in. We're doing things like CRT or perhaps ECMO and we'll also look at how to adjust antibiotics in those situations. The objective of today's presentation is to understand the impact of critical illness on antimicrobial pharmacokinetics and pharmacodynamics. As we have a more general audience today and it's not just targeting pharmacists, I'm going to go through some of the basics of pharmacokinetics and pharmacodynamics and some definitions of commonly used terminology that we'll be discussing today. Pharmacokinetics is considered to be the movement of drugs through the body. This is typically broken up into absorption, distribution, metabolism, and elimination. Pharmacodynamics is the result of the drug interacting with the target receptor or site of action. In other words, pharmacokinetics is what the body does to the drug and pharmacodynamics is what the drug does to the body. Protein binding is an area that has a lot of misconception around it. A drug that is protein bound is bound to plasma proteins, most commonly albumin. When we consider a drug that is highly protein bound, say for example, ceftriaxone, that's bound at about 90%, what that means is that when we give a dose of say a gram, 90% of that dose is bound to albumin in the plasma, does not travel to the site of action, and does not work at the target site. So drugs that are protein bound, it's not a mechanism of delivery of the drug to the site of action. When the drug is bound, it is neither active nor able to be eliminated. The volume distribution is the theoretical volume comparing the plasma concentration to the administered dose, and I'll give you a visual on this in a couple of slides. And finally, log P is a measure of how hydrophilic or lipophilic a substance is, and this really comes into play when we're considering things like ECMO, which we'll get into near the end of the presentation. Here we have a simplified version of the compartment model for drug distribution. We start in the bottom left with our patient receiving a medication. The two compartment model includes blood and tissues. When we first give a dose of a medication, the concentration rises from zero to the peak. Immediately following administration, the drug leaves the blood compartment through two mechanisms. First, the liver or kidneys begin metabolizing the compound, and second, the drug distributes into the tissues. During this distribution phase, we call this alpha-order elimination. When the concentration gradient between the blood and tissues is established, which depends on a lot of drug-specific factors, we then have removal of the drug from the tissues and steady-state elimination, represented in the graph at the bottom by the beta phase. We'll go through this concept in more detail on how it applies to patients throughout the presentation, but keep this modeling in mind as we revisit volume distribution. Now that we've covered what normally happens when you inject a drug into somebody, let's talk a little bit more about how the antibiotic is actually going to be working for us. In this graph here, we've got a concentration in micrograms per mil along the y-axis and time in hours along the x-axis. This dashed line across here represents the minimum inhibitory concentration that's necessary to kill bacteria, and our various classes of antibiotics interact concentration versus the MIC in a variety of ways. This line superimposed here is the example of giving a medication. We give the drug, the level starts at the bottom, goes to a peak, and then is gradually eliminated. Our first class of antibiotics are the time-dependent antibiotics. These antibiotics target a certain percentage of the time interval of the dose above the minimum inhibitory concentration. In a perfect world, we're targeting about 100% of the dosing interval, but if you look at the actual literature around time-dependent antibiotics, the typical target is somewhere between 40 and 60% of the dosing interval, having the concentration above the minimum inhibitory concentration. This is most commonly our beta-lactam antibiotics, and if we are finding that we're running short on this, we can either decrease the time between doses by shortening the interval, we can extend the infusion of the administration, or we can even do continuous infusion. For concentration-dependent antibiotics, these antibiotics restrict their effects based on the level of peak achieved, or C-max, above the MIC. This is primarily aminoglycosides, and if we've got a situation where we feel like the aminoglycosides not being as effective as it could, you can counteract this mainly by increasing the dose and increasing the dosing interval. Finally, the last sort of hodgepodge classification would be those antibiotics that have more of an AUC over MIC, or area under the curve above the MIC. This includes things like vancomycin, daptomycin, tetracyclines, and polymyxins. In order to optimize these medications, to optimize your AOC to MIC ratio, it's fairly variable adjustment. It kind of depends on the bacteria and the site of the infection, but doing things like increasing the dose and playing with the interval as well are also options. Thinking back to our two-compartment model, unfortunately for us, we don't actually have a giant square of all of our tissues and bones and everything else that's one amorphous, continuously circulating piece for antibiotics to penetrate into. Based on the relative vascularity and some other factors, antimicrobial distribution and penetration is different depending on the site you're trying to get to. In general, it's much harder to penetrate bone, lungs, and brain, as opposed to things like skin, other soft tissues, the bladder, and in the intra-abdominal space. So even if we have the perfect antibiotic with the perfect pharmacokinetic criteria, we still have to get enough of it to the site of action, which is challenging at the best of times. And to give you a little bit more of a visual on this, if we are giving a drug, let's use ceftriaxone as an example, and we give a two gram dose into the blood, a certain percentage of that will get into the tissues. But since ceftriaxone is a time-dependent antibiotic, we're trying to target that time above the minimum inhibitory concentration. And so if you have a perfectly adequate amount in the serum, you've got, let's say, 15% of that that actually reaches the lung tissue. And if that 15% that's actually in the lung tissue at the site of the pneumonia is not above the minimum inhibitory concentration for at least 50% of the dosing interval, you're going to have antibiotic failure. And so when we're considering all the pharmacokinetic and pharmacodynamic changes of critical illness, this is really the meat of what we're trying to fix. We're trying to get enough drug to the site of action for long enough to be efficacious in our infection. Before we get into the specifics of the changes of the body and how that impacts the medications, let's do a quick review of what our normal expectations should be for hydrophilic and lipophilic antibiotics. Hydrophilic antibiotics, like our beta-lactams, aminoglycosides, vancomycin, and lonazolid, are expected to have a relatively low volume of distribution. They're generally rarely cleared and they tend to have a lower penetration to things like the brain or perhaps the lungs. Lipophilic antibiotics, like our fluoroquinolones, macrolides, lincosamides, and tegacycline, are expected to, in general, have a higher volume of distribution. They're also generally hepatically cleared and tend to have a higher degree of penetration into various tissues and spaces. We'll now move into the physiologic changes that occur in the critically ill and have some general thoughts on what we can do to counteract these changes with changing our approach to the antimicrobial dosage. This slide represents the general expected physiologic alterations of a patient who's presenting to us in acute sepsis or septic shock. The volume of distribution is going to be massively increased because of the fluid shifting of sepsis. Protein binding is expected to significantly decrease since albumin, as an acute phase reactant, is consumed and decreases, which therefore increases the unbound drug that's available for clearance by the body. With early sepsis and early septic shock, when the cardiac output is acutely increased, this enhanced cardiac output increases drug delivery to elimination sites like the kidneys and liver. And so when we're considering something like augmented renal clearance, this is in part due to the increased cardiac output in early critical illness. Finally, for shock states, because of the alterations in capillary flow, we expect a decreased antibiotic penetration about five to ten times less than was expected in a normal physiologic state. So with all of these things together, we can see that a normal standard antibiotic dose and a normal standard antibiotic interval are likely not going to be enough for our critically ill patients. As a result of the physiologic changes that we observed in the previous slide, we now will expect a shift in the drug behavior from the normal that I still have on the left side of the slide over to the right side of the slide. For our hydrophilic antibiotics, again this is beta-lactams, aminoglycosides, vancomycin, and lonazolid, we'll see an increased volume of distribution, we'll see more rapid clearance of the antibiotics early in sepsis and septic shock, and we'll see a delayed clearance later on in the disease process. We'll also see worsened penetration, and this is partially because of that increased volume of distribution. If you go from a volume distribution of 0.5 liters per kilo up to 10 liters per kilo, there aren't enough molecules to create a gradient to distribute into the tissues as there would be when the volume of distribution was lower. For lipophilic antibiotics, again our fluoroquinolones, macrolides, lincosamides, and tegacycline, we actually will expect a relatively minimal change in volume distribution. Because the drugs are more attracted to fats than water, the large water volume shifting doesn't really impact their distribution quite as much. However, we still do see an early rapid clearance and a late delayed clearance of the medications, and we do also see worsened penetration and worsened ability to get to the site of action. We're first going to dive into increased volume of distribution and what we can do at the bedside to help counteract this problem. I'll propose to you a fairly standard situation that we deal with. A 50 kilogram patient presents to your ICU with community-acquired pneumonia. Because you know this patient is from the community and they're 50 kilograms, you decide that a reasonable dose would be a 1 gram of ceftriaxone. The drug will distribute, a certain percentage will get to the lungs, and begin killing bacteria. If we consider, however, the 50 kilogram patient presenting in sepsis, their effective volume of distribution is much larger. Therefore, the same size dose of ceftriaxone, 1 gram, will result in much less at the effective site of action in the case of septic shock, and in theory, worse outcomes. We have actually seen this play out in the literature. There's one study notably that identified worse clinical outcomes in ICU patients with community-acquired pneumonia given 1 gram versus 2 grams of ceftriaxone. So the bottom line here is to load hydrophilic antibiotics in the critically ill. A common misconception and a common thought would be to say, well this patient has chronic end-stage rheumatoid disease, they're on dialysis. My typical dose of, say, cefepine would be to give 500 milligrams every 24 hours. For the first dose in a patient with sepsis or with septic shock, that's not appropriate. We have to fill the tank before we can worry about removing medications. The ED is an area that's actually very good at doing this. They tend to be very aggressive in their initial dosing, which is wholly appropriate in the case of a patient with sepsis or septic shock. Let's next consider altered protein binding and examine the ways in which it impacts the medication efficacy, as well as what we could do about it at the bedside. There are three main challenges that we expect in patients that have hypoalbuminemia. Firstly, an increased volume of distribution, which is primarily going to impact hydrophilic drugs. Secondarily, because there's less protein for the drug to bind to, we expect an increase in overall clearance of the medication, so more drug will be delivered to the kidneys or to the liver for removal. And finally, we expect that there will be less free drug available to work. So in summary, low albumin means less actual drug that's able to kill the bacteria. Aldelmo Lins and colleagues published a really excellent review of antibiotics and protein binding in 2011 in Clinical Pharmacokinetics. What I've done here is I've replicated for you one of the tables of that paper, and I would just like to highlight a few things with the drugs that we use very commonly ICU. First, if you look over on the left here, you can see that drugs that are considered highly bound are more than 70% bound to albumin, and this includes ICU workhorses like ceftriaxone, erdapenem, doxycycline, and even things like our anticephalococcal penicillins, nafcillin, oxycillin, as well as cefazolin. For drugs that are moderately bound, this is that 30 to 70 percent range, we wouldn't expect necessarily that to have quite as much impact of low protein as we would at the highly bound medications, and this would be things like our fluoroquinolones, lenazelid, pipercillin, sulfamethoxazole, and vancomycin. And then finally on the right here, we have those drugs that are minimally bound to proteins. So in this case, low protein will really not impact your approach. This would be cefepime, dorepenem, imipenem, maripenem, and the aminoglycosides, as well as metronidazole. Now I can't tell you that there's a lot of strong, prospective, clearly validated data that tells us that we should be driven towards one class or another, or one classification or another, in a patient that we know is coming in with with low albumin. For example, I wouldn't look at this and tell you that every cirrhotic patient who you know will have an albumin of one should never get ceftriaxone. I think we can all agree that's a little bit extreme. However, over the next couple of slides, I will walk you through some current evidence about specifically erdepenem and ceftriaxone. That's something to just consider when you're looking at the patient as a whole, and what else is going on with their pharmacokinetics and pharmacodynamics. In 2015, Zusman and colleagues looked at 279 ICU patients divided into erdepenem versus imi or maripenem. The drug dosing was pretty standard, with erdepenem at 1 gram every 24 hours, and maripenem went 2 grams 3 times daily, and imipenem 500 milligrams 4 times daily, all given as short bolus infusions. The primary outcome was 30-day mortality based on a logistic regression analysis, and patients were also separated into high or normal or high albumin versus low albumin, with a cutoff of 2.5 grams per deciliter. They found that low albumin and erdepenem usage together was associated with increased mortality. Note that, as you may expect, increased age, increased Charleston score, increased SOFA score were also associated with mortality, and that patients that had low albumin but were given imipenem and maripenem did not have the same signal for increased mortality. We've seen similar data with ceftriaxone and hypoalbuminemia in the critically ill. However, in the current studies, there's not statistically significant increase in mortality, just a trend towards worsened outcomes. These are both small single-center studies, however, and so I wouldn't take this information and move forward thinking I shouldn't use erdepenem or ceftriaxone ever in a patient with low albumin, but it should be taken into context along with the other antibiotics that are highly protein bound, especially in the cases of the acutely critically ill patients. So what do we do with the patient in front of us, who we know is coming in with an albumin of 1.5? Again, I'm gonna hammer this home through the presentation. An adequate loading dose in the critically ill is essential. For patients on ceftriaxone, I personally have a relatively low threshold to increase to q12 hour dosing if their albumin is low and the patient is not stable at 48 hours, and I don't have another explanation. So there's not, you know, a suspected MRSA infection that we're not covering, or a suspected pseudomonas, or an ESBL. If you're going to use erdepenem in these patients, I would also consider more aggressive dosing of the erdepenem. However, an alternative would also be to change the antibiotic. So again, if you think about antibiotics that are low binding of protein, cefepime, meropenem, imipenem, and then piperacillin is more of a moderate binding, as opposed to the erdepenem and ceftriaxone. Again, I wouldn't inherently not use erdepenem or ceftriaxone in the critically ill that has hypoalbuminemia, especially given that it's an acute phasoreactant, and perhaps these patients aren't chronically malnourished, but it's something to consider if you have otherwise unexplained failure of your antibiotics in a situation where they should be efficacious. Augmented renal clearance represents a fairly significant challenge, both in terms of identification and in terms of what to actually do about it in a patient who's experiencing it. Augmented renal clearance is defined as an estimated current in clearance of above 130 mls per minute per meter squared. It's estimated to occur in 18 to 80 percent of critically ill patients, which is not a helpful range. Hefney and colleagues last year published an excellent meta analysis and review aimed at defining augmented renal clearance and looking really at what the risk factors for it are. They identified a pooled prevalence of about 39 percent of a mixed ICU population in general, but a 75 percent prevalence in the neuro ICU population, and a 58 percent prevalence in trauma. They also found that male sex and age under 50 years is more associated with the incidence of augmented renal clearance. It's also very important to note that the calculated creatinine clearance isn't adequate to truly identify. The highest quality studies support a measured creatinine clearance through six or 24 hour urine collection. This is obviously very impractical for a wide-scale application, but formal diagnosis cannot be made on measured serum creatinine alone and calculated creatinine clearance from that. Several scoring tools have been proposed, but none are universally helpful in a broad range of ICU patients. The Mikami study referenced here looked at about 245 cases of patients who had continuously measured urinary creatinine excretion and they found a prevalence in a mixed ICU population of about 33% of augmented renal clearance. They found that the median onset was five days from admission in most patients with the TBI median onset being about 48 hours. They also identified that the median duration was about was roughly five days. However, this median was from quite a large range up to several months for some of the patients with traumatic brain injury. But somewhere between that two to five day window for a timing of onset from admission appears to be a fairly reasonable estimate for when you can expect to see augmented renal clearance. So you may say all right this is all well and good. My young patients that are males that were in traumas with traumatic brain injury this is what I got to look out for. But how do I actually dose these antibiotics in a way that'll be effective so that we're not just ruining our ability to target attainment. In 2015, Hobbs and colleagues provided a really excellent review of the literature that looked at the impact of proposed dosage adjustments in augmented renal clearance and how you could expect that to change your percent of target attainment. We'll run through this by each specific drug. So starting at the top here with piperacillin you can see that their target was about 50% time above MIC. Remember in the perfect world we'll get a hundred percent time of MIC but for overall validation targets anywhere between 40 and 60 percent is considered reasonable. Their dose adjustment that they identified was using a 4.5 gram load and then giving 4.5 grams every six hours but infusing it over three hours. They found that percent target attainment here improved to 80% when this dose adjustment was made. One thing I'd like to note as we start talking about extended infusion times of beta-lactams is that if you're going to use an extended or a continuous infusion strategy the first dose needs to be a full dose load dose normally. So in the case of piperacillin-tazobactam it should be 4.5 grams given over 20 to 30 minutes. The reason being if you're going to give an antibiotic over three hours to infuse the whole 4.5 grams your ability to achieve that initial C max that initial peak is going to be much lower than if you actually give the load and then on subsequent doses give that extended interval. Moving to maripenem we've got a similar strategy again targeting that 50% time above MIC here giving a one gram load and then at one gram every eight hours again infusing it over three hours and this percent target attainment got us up to about 82% likelihood of attaining the target. For cefazolin again targeting 50% at time of MIC you can see here that giving the the drug every six hours versus every eight hours over a standard infusion time really didn't have too much change in the percent target attainment achieved. For some baseline reference standard cefazolin dosing would be one to two grams every eight hours. So personally I would probably target a Q6 hour strategy just based on the numbers here. For cefepime looking at two grams every eight or two grams every six hours again infused over three hours was able to move from 89% target attainment to 99% target attainment. And finally for daptomycin which is targeting an AUC over MIC as we increase the average daily dosing from six to ten milligrams per kilogram per day you are more likely to achieve that target attainment at time of MIC. It is important to note that some of these percent target attainments are based on the estimated creatinine clearance that patients are demonstrating. So in the case of piperacillin, tezobactam, and meropenem if they had a medium creatinine clearance of a hundred and six or close to that 130 you have better attainment. However there are some studies with patients that have reported creatinine clearances that are calculated or that are that are measured of above 200 mls per minute per meter squared. A patient that's got that much clearance none of these strategies are really going to be that effective and you would need to do something like considering a continuous infusion. So the patient in front of us that young patient that young male patient with a TBI who's in your trauma ICU and it's day three and his estimated creatinine clearance is somewhere in the hundred and seventy hundred and eighties. What do we do with this patient? The first thing is having a high index of suspicion in an appropriate population. So this patient that we're proposing here would be the person that would be experiencing augmented renal clearance. We want to optimize our pharmacokinetic targets with aggressive dosing. Use a large loading dose. Utilize extended infusion beta-lactams. Early therapeutic drug monitoring if it's available at your center so that you can respond to a low level or an inability to achieve a target. And finally changing to a non-renally eliminated agent may be your best option to actually achieve a target. Another common scenario where this comes up that's not just the beta-lactams would be in the case of vancomycin. It's not uncommon for patients in their early 20s and late teens to require vancomycin doses very aggressively 1500 milligrams 2 grams every 6 hours. And as we all know the more vancomycin that you're giving a patient in 24 hours increases the risk for acute kidney injury. So in those cases at our institution we will sometimes switch the patient preemptively over to say deptomycin or even lenazolid to try and get away from the rapid renal clearance that's happening with the vancomycin. Before we dig into the impact of devices on our PK and PD I'd like to briefly touch on how to adjust drug doses in our obese population. Meng and colleagues in pharmacotherapy every few years publish this really excellent comprehensive review of updates in the literature with antimicrobial dosing and obesity and I highly suggest that you investigate the paper. With patients that have obesity their increased fat relative to their muscle mass increases the volume of distribution but it's in a nonlinear fashion and it is also very dependent upon the characteristics of the drug itself. Drugs that are hydrophilic with an otherwise relatively small volume distribution may not be appropriate to dose based on the patient's total body weight as we wouldn't expect that the medication is going to distribute widely throughout all of the tissues anyway. There's also potential in these patients for increased renal clearance owing to increased renal mass. All of these variables can have a significant impact on volume distribution and clearance. Note that a typical weight that you'll see for dosing cut off in in quote-unquote obesity recommendations is above 120 kilograms. Though depending on the study this definition of obese can vary extremely widely. Another challenge when you're trying to translate the literature that's available for obesity dosing into the actual patient is that you'll have studies that have say the maximum weight of the patient included will be 130 kilograms which isn't very helpful when the patient in front of you is one that weighs 180 kilograms. So on the next slide I've got a few dosing strategies for how you can approach these patients. A weight-based dosing requires an adjustment to minimize toxicity for patients that are obese and there's there's sort of three concepts for drug dosing weights that we consider. There's the total body weight, there's the ideal body weight which is based on height, and then a baseline weight assessment of either male or female, and then finally there's an adjusted body weight which is a calculation that takes into account kind of somewhere in the middle. So we'll use these these strategies like I have in the table listed below to try and account for the increased volume and the increased tissue that needs to be penetrated but we're also trying to minimize toxicity. So for aminoglycosides, daptomycin, and trimethylamethoxazole, we should use an adjusted body weight in our weight-based dosing considerations. For these patients we can also modify our monitoring strategies. Vancomycin is especially challenging as in general we consider that we should be targeting total body weight for dosing and the asterisk that I have there on the slide is that we should also consider dose capping because if you think about all right I'll do my 25 milligram per kilogram load of vancomycin in my critically ill patient to fill the tank but the patient weighs 200 kilos you're proposing to give that patient 5 grams of vancomycin and we know that part of vancomycin toxicity is related to the total daily dose. So a modified approach or a dose capping approach of saying no load greater than say 2,500 milligrams or 3 grams can be taken in these patients to protect their kidneys. However that doesn't help you then decide on what the appropriate interval is going to be. So a strategy that we rely on at my institution is to give that load give an appropriate 2,500, 3 gram load and then you can obtain two levels after that load before the next dose to understand what is the individual's kinetics. This requires the use of some calculus, which your pharmacist will be very happy to do for you, but it's another option to try to individualize the patient's subsequent dosing and that leads nicely into the final point for obesity dosing which is therapeutic drug monitoring is essential if you can get it. Now that we've talked through the changes that the individual patient's body is actually going through their acute illness, let's shift a little bit into going through the impact that our devices and our machines have on our antimicrobial pharmacokinetics. When we consider continuous renal replacement or CRT and its removal of medications, it's important to acknowledge that CRT relies on convection as opposed to dialysis which relies on diffusion. CRT's impact on drugs is dependent upon four individual factors. Drugs that are low protein binding, in other words they have a higher free fraction of drug to be removed. Drugs that are hydrophilic with a low volume distribution and a weight less than 500 Daltons are all more likely to be removed by the machine. The weight classification perhaps isn't the most helpful but it's something to consider when you're looking at let's say a medication that doesn't have a lot of data around whether or not you should adjust on CRT. The rates of drug removal differ between dialysis and CRT and dialysis dosing cannot be relied upon in CRT. It's also contingent upon the modality of continuous renal replacement therapy. There's continuous venal venous hemofiltration which is the most common. There's CVVHD or hemodialysis. HDF is hemodiafiltration and finally there's there's slow low efficiency daily dialysis or SLED. SLED is more of a hybrid between dialysis and CRT and typically only occurs for 6 to 12 hours a day as opposed to the 24 hours of the more continuous treatments. Drug dosing considerations and SLED are much less reported on in the literature and it would it takes a lot more specific tweaking. You've got a patient or your hospital system is routinely using SLED I highly encourage you to talk with your pharmacist about appropriate adjustments. When we think about our continuous modalities some general principles are that a higher flow rate indicates that more drug is likely to be removed. Your typical flow rate in septic shock is going to be somewhere between 25 and 35 mils per kilo per hour but there are times when lower flow rates are used and in those times again you will not need to give nearly as much medication because you won't expect as much to be removed. For our primary classes that we utilize beta-lactams and vancomycin are going to be at a higher risk of removal. Finally talk to your pharmacist. Talk to your pharmacist. Talk to your pharmacist. We have access to a lot of really excellent resources and review articles and literature about with drug monitoring and TDM to give you the best possible option for dosing with as much evidence as exists. With the increase in the use of extracorporeal membrane oxygenation in the last five to ten years in critical care the need to understand the impact of the circuit on medication variables has increased significantly as well. Fortunately there are a lot of excellent teams of researchers out there that are working on understanding the most commonly used medications and how the ECMO circuit impacts them. Here we have the return of the compartment model just adding the ECMO compartment. As you recall we've got giving a drug into the blood which then distributes into the tissue back and is eliminated. However when the patient goes on circuit we're adding a significant volume with the circuit tubing as well as the pump. So the first change that we expect when we put a patient on circuit is a vastly increased amount of volume of distribution. This is primarily going to impact medications that otherwise have a low volume of distribution because the effect of circulating volume of the drug significantly decreases. It's fairly common for us to give another loading dose of medication in a patient who's on a steady dose of a drug that is affected by ECMO. Additionally the ECMO modality can also have an impact on drug clearance. VV ECMO tends to result in more normal feeling pulsatile blood flow whereas VA ECMO at very high flow rates can produce a non-pulsatile flow. If a patient is experiencing non-pulsatile flow this can alter the perfusion of tissues and reduce capillary circulation. The kidneys can interpret this pulse as blood flow as hypotension and then activate RAS which results in lower urine output and lower overall filtration of medications. Monitoring of these medications particularly metaglycosides when patients start ECMO is extremely important as you may require an overall higher dose but an increased interval so that you're not causing a buildup in some toxicity. The other complication with ECMO is the composition of the circuit itself. The polymethylpentene fiber tubing and the pump materials can sequester medications. Medications that are highly protein bound and those that have a high log P are the medications that are at risk of this complication. As you might recall log P is a measure of the relative hydro or lipophilicity of a medication and the higher the log P the more likely the medication is to be lipophilic. So what happens here is the drug actually binds to the tubing itself and therefore is not effective to act. The complicating factor is that even after medication administration has ceased drug that was sequestered to the tubing can be released from the membrane and continue to cause prolonged effects. This tends to be more of an issue with sedatives than antimicrobial agents but it is a possibility. The paper reference here is one I wrote last year with some of my fellow ICU pharmacist colleagues. It was a comprehensive literature search and summation of the ECMO impact on medications. I've pulled out a summary of just the antibiotic findings that we had here along with what our general suggestions were for empiric dose changes. It will be a little excessive to go through individually so I'd just like to give you a few highlights. The first consideration was is the medication sequestered and does that change our dose approach? The second question was is the volume of distribution impacted and does that change the dose approach? So the way that this table is set up would be the PK characteristics of the medication, whether or not the circuit actually takes a hold of it, and then what to change. And what you see as you skim over this table is that when we're considering a lot of these medications that are not highly protein bound, that are not lipophilic, that there's really not much empiric adjustment that we have to do because of the ECMO circuit once we've accounted for that increased volume of distribution. A lot of these we would only adjust if the patient was on concomitant CRT. However, things like vancomycin and piperacillin-tased Vactam may require initial higher load and higher maintenance dosing and potentially extended or continuous infusion, especially if you're dealing with an organism that has higher minimum inhibitory concentrations. To summarize, with ECMO an increased volume of distribution may require higher loading and higher doses of those concentration-dependent antibiotics that are relying on a C-max above the MIC target, like aminoglycosides. We should consider therapeutic drug monitoring when it's available for things like the beta-lactams, and we should utilize aggressive dosing in patients that are also requiring CRT. Additionally, you should talk to your pharmacist about any dose changes that need to be made when you're putting a patient on circuit who's also on antibiotics. The use of molecular adsorbent recirculating systems are not very common, but because of that, when they are utilized it becomes a little bit of a scramble to figure out what is the right way to adjust medications. MARS is typically utilized in patients with acute liver failure as sort of a quote-unquote liver dialysis. It's adjunctive to a CRT circuit, so many of the medication considerations are going to be dependent upon their CRT characteristics that we already covered a little bit earlier. Drugs with low protein binding or a high volume distribution we generally will treat as those standard CRT dose adjustments. However, in drugs that have moderate to high protein binding, so this is anywhere between 30 and above 70 percent protein binding, and a small to moderate volume distribution can be impacted and removed by the MARS therapy. So when we're considering how to approach this, again there's not what you would call a lot of strong, clear, duplicative, replicated evidence for how to adjust. So as we're considering the pharmacodynamic properties of these medications, the general thought on how you can approach dosing I have listed in the table here. So when we think about our time-dependent antibiotics, again this is going to be things like beta-lactams, our approach is to bolus prior to starting them on MARS, to consider using continuous or extended infusions, and to increase the frequency of dosing. For drugs that are in that AUC over MIC category, like vancomycin, we'll increase the dose and frequency, and for concentration-dependent antibiotics, like our immunoglycosides, we'll increase the dose. Therapeutic drug monitoring is an extremely important component to be able to actually understand what's happening with the medications and respond if there's a problem either leading towards toxicity or towards lack of therapeutic efficacy. I hope we've taken away some of the main following points from my presentation today. Critical illness has a significant impact on the pharmacokinetics and pharmacodynamics of antimicrobial therapy. Our volume of distribution, protein binding, augmented renal clearance, antimicrobial penetration, are all affected just by the act of the patient being sick, and that's before we add in all of our other devices. In general, with critically ill patients, aggressive initial dosing should be utilized with a patient-focused reassessment on subsequent adjustments of both interval and following doses. And finally, talk to your pharmacist. We love this stuff. I could have spent an hour presentation to just pharmacists on just a few of the slides, so please talk to us. We're always happy to help. We're happy to give you the pages and pages and pages of literature that we have for why we're really excited about changing a dose, so please use this as a reference when you're dealing with critically ill patients and concerned for changes in PK and PD.

critical illness

antimicrobial pharmacokinetics

pharmacodynamics

protein binding

volume distribution

dosing antibiotics

continuous renal replacement therapy

pharmacists