Thank you so much for the opportunity to be here in the master class today on saving the kidneys. In this section, we are going to be talking about patients with acute kidney injury who have needs for certain medications and nutrition. I'm Erin Barreto. I'm an associate professor of medicine and pharmacy at the Mayo Clinic in Rochester. I'm really happy to be here with this excellent panel. These are my disclosures for today, and I have no relevant conflicts of interest for the material that we will be presenting. When we think about tackling medication and nutrition therapy management in acute kidney injury, it's a really complex situation. It feels to me at the bedside like it's constantly a moving target and there just aren't a lot of right answers, and the forever treat your patient, if you will, type strategy seems to lack a lot of specificity. This is especially complicated because acute kidney injury is highly prevalent in that estimates indicate that we're talking about one in five hospitalized patients here. All of these patients need medications and they need nutrition. I'm hoping by the end of this talk, we'll have given you some new evidence and some tips for how we can approach clinical decision-making that is a bit more specific and tailored for our patients in this sticky situation. We have three core objectives for today's lecture. First, we will discuss the pharmacokinetic and pharmacodynamic management of medications in the presence of acute kidney injury. And then I'm going to present the concept of nephrotoxin stewardship, which is a bit of an emerging term, but uses some simple ideas that we're all probably pretty familiar with. And then last, we'll talk about the importance of providing appropriate nutrition to critically ill patients with acute kidney injury. Medications commonly eliminated in nephrotoxic drugs are nearly ubiquitous in the critical care setting. Up to two-thirds of the medications that we use are renally cleared and about a quarter are nephrotoxic. And I'm going to focus today using the example of antimicrobials because it is some of the most well-evidenced based, but you can think of other drugs that are relevant to your patient population that might fit into one of these categories. For example, antithrombotics have narrow therapeutic windows like bivalrudin or low molecular weight heparins. They would be considered renally eliminated. And then there's of course countless other drugs that are nephrotoxic, including NSAIDs and calcineurin inhibitors and other things like that. If we focus though on antimicrobials, there are many renally eliminated antimicrobials and many nephrotoxic antimicrobials, but very few that are just one or the other. The majority kind of live at this intersection. So for example, beta-lactams are small molecules that are hydrophilic and very renally eliminated, but we know that they can be associated with acute interstitial nephritis. Similarly, so if we look at something like antivirals like acyclovir, that would be used at high doses for meningoencephalitis. This would be again an example of a renally eliminated and nephrotoxic antimicrobial. If we focus in on the renally eliminated piece of the puzzle here, the idea here is that when we dose medications based on kidney function, the assumption is that it's based on steady state. Essentially, you want to be standing on level footing to be able to determine what dose and interval to use, how many grams on once a day, twice a day, three times a day, et cetera. Yet when we have unstable kidney function, such as an acute kidney injury, our tests and tools don't work correctly. I was trying to think of analogies to represent this. It's kind of like looking through a rear view mirror or driving without a speedometer. You really can't make a plan without stable or consistent information, so it always feels like hindsight is 20-20. If we think about estimated creatinine clearance or estimated GFR, it doesn't really matter what the calculation is if the patient is in URIC, because even if the EGFR says that it's 40 mL per minute, if the patient is making no urine, they're not clearing anything, creatinine, drugs or otherwise. So it really doesn't matter what the calculation says. I think we all know this clinically, but the problem is there's no great way to quantify it in that situation to make a better decision. So I'm going to propose two strategies today that might be amenable to bedside application at your center that might put a bit of a finer point on the choices that we make in practice, as opposed to just our clinical gestalt. So the first of those two strategies is the kinetic EGFR, and over the years there's been a variety of different equations that have been developed to try and account for unstable kidney function but still use creatinine, and the pioneer for this has been Roger Jelliffe. But the most recent equation that has been gaining some traction is that from Dr. Sheldon Chen, published in Jason in 2013. So Chen uses an algebraic equation that kind of has two pieces. The first half of the equation shown here at the top of the slide on the left is what we think of kind of as the static part of the equation. This is similar to the UV over P equation that people might be familiar with in nephrology, where it kind of accounts for the production rate and the average creatinine that a patient has. On the right is the more dynamic component, so this accounts for the changing creatinine over the change in time. So if you look at the graph here just to kind of visually represent it, on the left y-axis we've got the serum creatinine, and on the right y-axis we've got a representation of kidney clearance in GFR, and then on the bottom we've got time. So what you can see is the serum creatinine kind of is sharply rising, but then plateaus around day three. If you use the standard clearance equation shown in the orange boxes, there's a constant downward slope that you're seeing here, meaning that it suggests to the clinician that their clearance is continuously falling. However, if you use the KEGFR equation, what you see is that rather than this perpetual downslope in the regular clearance, you actually see an upward inflection beginning as early as day one. So this is represented by the gray diamonds, and what this is telling us is that the KEGFR is detecting a slower rate of rise in the serum creatinine. So it more carefully shows that this plateau is probably a period of time where the patient actually is developing a slight improvement in their EGFR, but the creatinine hasn't fully demonstrated that yet. Now there hasn't been a lot of studies on the application of kinetic EGFR to drug dosing, but there is this article from Dr. Pai at University of Michigan, and he looked at a data set of around 2,500 patients treated with vancomycin over about five years, and he performed a population pharmacokinetic analysis to see if altering the kidney clearance equation meaningfully affected their model predictions. Patients were grouped according to the stability of their kidney function. So you can see here on the y-axis, we're looking at the change in serum creatinine, and you can see everybody was put into one of four groups. Do they have a stable kidney function? Did they have a declining serum creatinine or improving kidney function? Did they have a rising serum creatinine or worsening kidney function, or were they kind of a mixed picture? Each of these estimates was then included in a population pharmacokinetic model to predict clearance of vancomycin, and it was modeled either as time-invariant for patients, meaning their serum creatinine was effectively fixed or their EGFR was fixed at the first time point when it was available, or time-varying, where they were grouped according to those four buckets that I told you about earlier, worsening or improving or stable or mixed. What you can see looking at the AICs is that the Chen model, which is the kinetic EGFR that we've been talking about here, has the lowest AIC and the greatest negative delta AIC. This would suggest that this is the best-performing model relative to all of these other ones that were evaluated. It accounts for unstable kidney function and the change over time, as opposed to just fixing it at a single time point for patients. This is promising, although a bit more theoretical, I do think that it could be applied to the bedside for how we approach medication dosing and management. The benefit of the KE-GFR is that it uses an existing tool in serum creatinine, which is widely available and doesn't require a lot of sophisticated math to evaluate the KE-GFR. Rather, you can just plug it into an online calculator. Another alternative is to use Cystatin-C. Cystatin-C is an alternative functional biomarker that can be used in place of or in addition to serum creatinine to assess the kidney function. It's a low molecular weight protein that's produced by all nucleated cells. It's eliminated by a glomerular filtration, and then it's reabsorbed and catabolized in the proximal convoluted tubule, so there's no real meaningful systemic reabsorption. It's easily measured in the serum and urine. We're going to focus on the serum today. The urine's a bit of a different situation that's a bit beyond the scope. It's relatively inexpensive, although in fairness, it is more expensive than serum creatinine. Cystatin-C has been evaluated for medication dosing and monitoring. The systematic review that we conducted through 2017 revealed 34 studies with 16 different medications, most of which are antimicrobials, but also anti-neoplastics and cardiovascular medications. EGFR cystatin-C was assessed using different equations. At that time, the CKDF equations were generally not available for the most part yet, and the Cockroft-Galt equation was the primary comparator, which I think still, for better or for worse, is used in a lot of centers. What we demonstrated in this systematic review was that EGFR cystatin-C predicted drug clearance in pharmacokinetic studies at least as well, if not better than EGFR creatinine, but rarely were the biomarkers combined, which has been suggested to try and overcome the non-renal determinants of either of them, and infrequently were targeted attainment and clinical outcomes evaluated. These are mostly PK studies. We aimed to build upon these data and evaluate the clinical application of a cystatin-C-based strategy to dose and monitor medications in the ICU. Through a series of studies, we developed a new nomogram for vancomycin dosing that included the EGFR creatinine cystatin-C, using the CKDF equation, expressed in mLs per minute, which is not the native unit. This is a re-expression. This is just an excerpt of the nomogram. You can see that for a given patient, you have to look at their weight, their EGFR, and at that time, we were using trough, so what their goal target was, and a recommended dose. For example, for a patient who's 80 kilograms with pneumonia, with an EGFR of 65 mLs per minute, they would get a dose of 1,200 milligrams Q12. We then went on to evaluate the performance of this new nomogram when applied to three critical care practices relative to our standard dosing strategy using Cockroft-Galt creatinine clearance in historical controls. What you can see represented on this graph is the percent of target achievement. This is individual target achievement in patients in the historical controls compared to the cystatin-C guided dosing nomogram. What I hope you can appreciate is that there was a significant improvement from 28 percent target achievement to 50 percent target achievement with the use of the new nomogram, and it did not come at the expense of a greater proportion of supra-therapeutic concentrations. This was really a reassuring finding and suggests that, indeed, you can develop and implement cystatin-C-based strategies to improve pharmacokinetic and pharmacodynamic attainment. Clinical outcomes remain to be demonstrated in this data or these data or in others. Now, importantly, in that previous literature that I showed you, that was not unstable kidney function. That was stable kidney function using cystatin-C, and to the best of my knowledge, there is not data for medication management in unstable kidney function using cystatin-C, but I think this study can be a very helpful one. These data come from Nejat and colleagues in Nephrology, Dialysis, and Transplantation. What they show here is a serial evaluation of creatinine and cystatin-C in critically ill patients, and they looked at which biomarker rose first, essentially. Everybody had both biomarkers checked daily for up to seven days, and then they looked at which one rose first. Now, there was about three-quarters of the patients out of the 318 that were included that didn't have any rise, so what I'm showing is the remaining about one-fourth of patients. On the left is the patients who had a serum creatinine rise first. You can see it was a minority, only 13%, and even then, the curves for creatinine and the curves for cystatin-C are kind of superimposable. On the right are the patients who had both biomarkers rise effectively simultaneously, but the bulk of the patients that did have a rise demonstrated that the serum cystatin-C, represented in the orange boxes in the middle here, rose earlier than the serum creatinine. So what this can tell me is that cystatin-C may have a more favorable kinetic profile in patients who have unstable kidney function with acute kidney injury. So that could be promising, as we know that creatinine typically has an up to 48-hour lag from the onset of kidney dysfunction before we even see a rise, and it's heavily affected by nonrenal determinants, including things like skeletal muscle mass, deconditioning, malnutrition. So for a lot of reasons, this may indicate that cystatin-C is a more timely marker of kidney deterioration. Now, I'm not aware of any data that looks at cystatin-C specifically in the KEGFR equation, but this might be another way to kind of bring these together. You could theoretically do a KEGFR with cystatin-C instead of creatinine, and you might be able to then apply that directly to medication dosing with even better accuracy. So we just talked about the strategies to try and approach readily eliminated dosing a little differently. Let's shift gears and talk about drug-associated acute kidney injury or nephrotoxicity. About 23% of the drugs used in hospitalized patients are nephrotoxic, and we know that among the AKI cases in the hospital, nephrotoxins are associated with about one-third of them. And unlike sepsis or comorbid conditions or age, nephrotoxic medications are one of the few modifiable risk factors for AKI. There are lots of ways to classify drug-associated acute kidney injury or nephrotoxicity, and one of those proposed by Ravi Mehta and his group uses the clinical pattern and uses three different classification terms. One is the phenotype of kidney disease, one is the time course of injury, and one is the mechanism. With respect to phenotypes, it's not uncommon to see patients who have altered intraclamarular hemodynamics in the ICU setting. Anything that affects the effect of arterial blood volume will do this. So drugs that have an interface at that location, specifically we think of RAS inhibitors or NSAIDs or even loop diuretics, can certainly be associated with toxicity. There's also direct tubular toxicity like ATN, acute tubular interstitial nephritis, and then other slightly less common nephrotoxic patterns in the ICU. Nephrotoxicity can also be classified according to the time of onset, so it can be seen as acute, subacute, or chronic. And then the mechanism of injury relates to the dose dependence, so it can be a type A dose-dependent injury or something that's a bit more idiosyncratic like a type B. So if we think, for example, about aminoglycosides in this context, they would have direct tubular toxicity with an acute onset and typically a type A pattern, where perhaps if you decrease the dose, you might be able to get away with continued use. Another way to classify nephrotoxicity is based on the biomarker pattern. So when we think of creatinine in urine output, these would be considered functional markers as shown on the left of this matrix, but there are also injury biomarkers, things that suggest actual direct damage to the tubules. So a drug that causes neither dysfunction nor injury, obviously, is the ideal circumstance, but there are drugs that cause evident dysfunction but no actual injury. So an example of this would be a RAS inhibitor, where in the chronic setting, you see an acute increase in serum creatinine and a decrease in EGFR, but they're actually nephroprotective and not injurious in that stable chronic kidney disease sort of environment. In contrast, we've got drugs that may be associated with injury but early evidence of damage that hasn't fully progressed to see a decline in EGFR or something that is detected by serum creatinine. So this has been referred to as subclinical AKI. This is the upper right quadrant here. So this is another way that you could classify nephrotoxicity. And still yet a third way to classify nephrotoxicity is based on the cumulative burden. So this could be represented kind of by the concept of the triple whammy. So I already made mention of these drugs, but this was a large study that looked at patients who were treated with antihypertensive medications who experienced an episode of acute kidney injury during their hospitalization or those who didn't. And it demonstrated that increasing exposure to three medications that have activity in the same location, in this case, namely diuretics, a RAS inhibitor, and an NSAID, was associated with a significant increase in the risk for acute kidney injury. So again, total nephrotoxin burden, meaning the number of drugs or the intensity of nephrotoxin exposure is also another way to classify it and worth getting into. Given the concerns about nephrotoxins, the term nephrotoxin stewardship has been coined. And we typically think of stewardship in the context of antimicrobials, which taken from the guidelines is effectively a coordinated set of interventions to improve and measure the appropriate use of a drug class and to select the correct drug regimen, including dose duration and route. And there's a lot of tactics that are used for nephrotoxin, or for antimicrobial stewardship, excuse me, that can be adapted to the nephrotoxin world. Sandy King-Gill has really pioneered this concept of nephrotoxin stewardship. And in a very nice summary paper published in Critical Care Clinics, she proposes three key goals of nephrotoxin stewardship, including coordinated strategies to enhance medication safety, ensure kidney health, and avoid unnecessary costs. One very successful nephrotoxin stewardship program is the NINJA Program, or Nephrotoxic Injury Negated in Just-In-Time Action. This was an initiative started at Cincinnati Children's, but has since expanded broadly to include non-critically ill pediatric patients. It's a little different population than ours in the ICU, but still very relevant. A nephrotoxin alert was developed based on a standard list, and simply anybody who had three or more nephrotoxic medications on their list received a pop-up for a pharmacist to review. And this dedicated pharmacist for the NINJA Program then would look through the cases and recommend to the team that the patient undergo daily creatinine monitoring. So this would be patients who have a high cumulative nephrotoxin burden defined as three or more exposures, and all that the pharmacist was doing was asking for creatinine to be monitored. The rest of the decisions about drug substitution, or drug levels, or whatever, that was really at the discretion of the care team. And then they looked to see whether people adhered to this protocol and what their nephrotoxin exposure and AKI rates were. So a relatively simple intervention, right? But there were some significantly favorable results. What you're seeing here on the y-axis is the nephrotoxin exposure rate with the mean represented in the orange bar, and then time represented on the x-axis. So what I can hope you can appreciate is over the course of three to four years in this project, there was a steady and significant decrease in the nephrotoxin exposure rate. And this was associated with a corresponding decrease in the incidence of acute kidney injury and the intensity of acute kidney injury, or how severe the stages were. So at three years, what they demonstrated was that there was a 31% decrease in the intensity of acute kidney injury and a 64% decrease in the rate of acute kidney injury. And remember, this wasn't something where they were trying to per se decrease the nephrotoxic medications with a certain intervention. They were just asking the teams to check a serum creatinine. Now, obviously, in adult critical care, it's pretty common to have a serum creatinine evaluated. But you can imagine that this simple idea of looking at your cumulative nephrotoxin burden could trigger conversations at the team level about how to more safely use them in patients. Indeed, in this study of adult critically ill patients at UPMC, there was an evaluation that looked at cumulative nephrotoxin burden. And in those who had a high burden, they performed the urinary TINP2-IGFPP7 test, or this is the test that is commercially marketed as nephrocheck, which is intended to be a early warning sign of potential future acute kidney injury. And among the patients who had a positive test, what I hope you can appreciate is that they had a significant improvement in the proportion of patients who had a decrease in their exposure to nephrotoxins or contrast. They had patients who had better volume evaluations. And in those who were on renally eliminated medications, they were able to undergo dose adjustment or therapeutic drug monitoring. So this is a way that the NINJA concept with nephrotoxin stewardship could be applied to an adult critical care population. To wrap up this section on medication management before jumping into nutrition therapy, I thought I would just summarize the conversation points with the patient case. So this is a 67-year-old admitted with hypoxemic respiratory failure and shock, thought secondary to pneumonia. The patient is intubated, sedated, and on pressors, and they have an elevated serum creatinine and oliguria. If you look at their KEGFR, what I hope you can appreciate using the graphs on the right is that the KEGFR, which is in the bottom panel, is starting to show an upward inflection at that kind of 48-hour mark, whereas with the creatinine clearance, you're still seeing deterioration. So you could see how this would be an opportunity then to catch a patient who's experiencing renal recovery and optimize their drug doses earlier. Again, it's hard to detect the renal recovery when looking at the blue serum creatinine pattern, but I think we would all agree it does look like it's plateauing. So this is really the time where we want to be starting to ask the question about increasing drug doses of renally eliminated medications slightly. If you have access to cystatin C at your center, you could check it in this situation, and what you would see is that the cystatin C concentration is two milligrams per liter, which would be elevated. The numbers for cystatin C and the numbers for creatinine look similar, although the units aren't different. So one is essentially a good cystatin C. In patients who have a stable kidney function and you mistrust serum creatinine, let's pick a patient who has ALS or has bilateral amputations. Maybe a cystatin C-based dosing using a nomogram such as the one I presented for vancomycin could be considered, but in the context of unstable kidney function, you could explore the use of cystatin C if you had enough serial measurements of it in the KEGFR equation to guide dosing. Also in this patient, it's evident that they have acute kidney injury, so nephrotoxin stewardship is of paramount importance to optimize their chances for recovery without complication. So considering the use of the KDGO bundle, this is a prevention bundle outlined in the guidelines that talks about ways to increase the chances that the patient does not go on to have worse outcomes after their AKI. Use of a multidisciplinary team to assess their cumulative nephrotoxin burden, and if you have access to novel damage biomarkers like the one I showed in the UPMC model, that would be something you could consider as well. Shifting gears then slightly away from medication therapy, let's talk about nutrition therapy in acute kidney injury. Well, nutrition is a very complicated teeter-totter in acute kidney injury in that malnutrition has been documented in up to 40% of AKI patients, and it's associated with increased risk of mortality and greater complications, but it's also well known that overfeeding has its own issues, including dysglycemia, which is already common in the ICU, hepatic issues, lipid abnormalities. So we're kind of trying to balance ourselves on this seesaw between under and overfeeding over the course of dynamic critical illness. Looking across the various guidelines that have been published on the energy requirements in critically ill patients with acute kidney injury, you can see that there's a variety of recommendations. The Europeans heavily encourage the use of indirect calorimetry, but we recognize that in critically ill patients, this may not be available or reliable, and therefore we can take cues from their recommended energy goals of around 20 to 30 kilocalories per kilogram per day for critically ill patients with acute kidney injury. It is important also to indicate that in the ASPEN SCCM guidelines that patients with acute kidney injury are recommended to be on a standard formulation of nutrition support as opposed to a specialty formulation, and it is recommended that they follow standard recommendations for protein of 1.2 to 2 grams per kilogram per day. Another mechanism to estimate a patient's energy requirements is through the use of estimating equations like the Harris-Benedict equation. You can see in this study where patients with stage three acute kidney injury who require nutrition therapy were included, they performed indirect calorimetry every 48 hours and then compared it to the equation-calculated energy requirements as well as the proportion that the patient actually was prescribed and received. In the table, you can see that regardless of whether the patient was or was not on mechanical ventilation, the proportion of calories prescribed was and received was around 200 and 300 kilocalories per day respectively lower than that which was suggested by the indirect calorimetry. As you can see by the Bland-Ultman plots, relative to indirect calorimetry, the equations as well as the fixed calculation using 25 kilocalories per kilogram per day had quite a spread of data around the line of no difference which would be the horizontal line at zero. This suggests that in all cases you had both the risk for overestimation and underestimation of the patient's caloric needs although depending on the equation it would change the likelihood of which of those was more consequential. Now we've been talking a lot about energy requirements and protein requirements in critically ill patients with AKI but I would be remiss to not mention fluid as well. This is just one set of data from Swedish hospitals that looks at fluid administered to critically ill patients on days three through seven of their ICU stay so largely out of the resuscitative phase. You can see that patients receive on average 800 to a liter a day of medication related fluid in addition to their nutrition and resuscitation. This was corroborated by some data from the U.S. that showed that more than two to four liters of medication is diluent and fluid is administered to patients during their ICU stay up to a week. So certainly this is a major contributor and an opportunity to look at the sodium contribution, the dextrose contribution, and the overall fluid balance that is coming from multiple potential sources. Okay, wrapping up the nutrition therapy considerations. It is not easy to determine the energy requirements for a critically ill patient let alone one with acute kidney injury. Some of the guidelines recommend the use of indirect calorimetry where possible although in certain cases this won't be feasible or accessible and in those cases we will use oftentimes an estimating equation but it's important to know that the delivered requirements and the prescribed requirements are often slightly lower than what the indirect calorimetry would recommend. When we look at the performance of these estimating equations it can be highly variable. Patients can have both over and under estimation. As far as protein requirements, 1.2 to 2 grams per kilogram of protein per day is recommended for acute kidney injury not requiring kidney replacement therapy and you rarely need to use specialty amino acid solutions. It's more appropriate to just stick with the standard formulation and last it's important to be mindful of sneaky sources of fluid, sodium, and dextrose for that matter. As I demonstrated medication diluents can be particularly consequential but also looking at things like maintenance fluids is important. We've talked a lot today about patients with acute kidney injury and the complexities of the use of medication and nutrition therapy. I hope you took away some new information that you can apply at the bedside but I'd be happy to entertain any questions so please do feel free to reach out to me either by email or by twitter and I'd be happy to discuss things further. Thank you.

acute kidney injury

medication management

nutrition therapy

kinetic estimated glomerular filtration rate

cystatin-C

nephrotoxin stewardship

energy requirements