Hi, everyone. Thank you so much for the introduction and to the organizers of this event for the opportunity to speak today. These are my credentials, and I have no significant disclosures to report. In this talk, I'll be covering the use of AKI biomarkers that should be readily available at most institutions here in the United States, and how these biomarkers may be used to improve the diagnosis and care of patients with AKI. With that framework in mind, our learning objectives for this talk are to review readily available biomarkers that may be used to improve the precision of AKI diagnosis, and to review how improved diagnostic precision of AKI can be used to guide treatment and improve outcomes. A few guiding principles for this discussion. While the term biomarker, when discussing AKI, often refers to novel biomarkers, almost none of these are widely available for clinical use, and thus out of the scope of this talk, although I will cover more on that in my next lecture. Instead, we will be discussing other readily available biomarkers of AKI that can be leveraged to improve diagnostic precision and inform patient care. More importantly, the premise of both of my talks today is that not all AKI is created equal. It is a heterogeneous syndrome, and we need to do better to identify and implement tools to refine the AKI diagnosis. Relying solely on static serum creatinine and or urine output-based diagnostic criteria is problematic in both adults and children for a variety of reasons that I'll outline in this discussion. And then finally, you'll hear the term AKI phenotype throughout both of my talks today, and for the purposes of this discussion, we will utilize this term to describe a unique identifiable subset of patients with AKI that has associated prognostic and or therapeutic implications. As an intensivist, I love the analogy Raj Basu has employed of comparing real-time dynamic AKI phenotyping to the blood gas assessment of respiratory function. As you can see, many of the proposed biomarkers included here should be readily available, and there is emerging literature to support separation of AKI into clinically relevant phenotypes using just these data. As a reminder, these are the current diagnostic criteria for AKI based on serum creatinine and urine output, although there is expert consensus from a recent AdKey meeting suggesting the inclusion of tubular injury biomarkers, which we'll cover more in the next talk. There are so many issues with this diagnostic framework that could be an entire lecture in and of itself. However, the goal of this talk will be to discuss ways to identify AKI phenotypes with different prognostic and therapeutic implications using currently available tools, including serum creatinine and urine output. I'd like to start with urine output, and specifically oliguria as a biomarker. One of my mentors, Stu Goldstein, published this review in 2020 summarizing data in both adults and children that demonstrate that AKI defined by urine output criteria both captures a subset of patients who would be missed by serum creatinine alone and identifies a subset of patients who have worse outcomes. Unfortunately, despite this, few studies have utilized urine output criteria for AKI diagnosis, likely due to difficulty ensuring accuracy of these data on EHR extraction. So what are the data? In 2015, John Kellum and colleagues examined 32,000 patients treated over an eight-year period and classified them by serum creatinine and or urine output criteria to assess for AKI. As you can see, just over 14,000 patients had maximum AKI severity defined by urine output criteria, compared to about 5,000 by only serum creatinine criteria and 5,000 by both. Those who had maximum AKI stage defined by both criteria had uniformly worse outcomes, as highlighted by the red box. Additionally, outcomes between the serum creatinine only and urine output stage criteria were not hugely different, as shown in the first two columns. In a similar study by Bianchi and colleagues of over 15,000 adults over a 10-year period from 2010 to 2020, they demonstrated that consideration of urine output criteria compared to the isolated use of serum creatinine criteria enabled the identification of an additional 5,600 patients with AKI, and this represented a third of all patients in the cohort with AKI. Additionally, after adjustment for serum creatinine and comorbidities, urine output stages three, two, and three AKI were associated with higher 90-day mortality compared to the rest of the cohort. More recently, a group performed a multicenter retrospective analysis of over 6,000 adults undergoing cardiac surgery to assess whether or not oliguria had an impact on the epidemiology of AKI and outcomes. They found that isolated oliguria was very common, with almost 43% of patients impacted, and substantially more common than isolated serum creatinine rise following surgery. Not pictured here, 75% of patients in the entire cohort had oliguria meeting urine output criteria for AKI. Importantly, even patients with mild, so that is stage one AKI by oliguria, had worse in long-term outcomes at six months, which were mainly driven by the development of CKD. Those with both serum creatinine and urine output-defined AKI had even worse outcomes, driven by the development of CKD and mortality. These same trends are seen in critically ill children. A post-hoc analysis of the Multinational WEAR study performed in roughly 3,000 children with complete serum creatinine and urine output data demonstrated that children with isolated serum creatinine or isolated urine output-defined severe AKI had similar rates of mortality and significantly lower survival than those without severe AKI. Similar to the adult data, children with severe AKI by both urine output and serum creatinine had significantly higher mortality than either alone. Importantly, two-thirds of children meeting urine output criteria for AKI would have been missed by serum creatinine alone in this cohort. Furthermore, in this same cohort of children, those with severe AKI by urine output or serum creatinine and urine output both had significantly higher need for renal replacement therapy than those with severe AKI by serum creatinine alone, as represented by the relative risk shown at the bottom of this slide. In another secondary analysis of a WEAR, investigators developed the AKI score, combining the AKI stage from 0 to 3 based on serum creatinine plus the AKI stage from 0 to 3 by urine output criteria to get a possible score of 0 to 6. As demonstrated in this table, in general, increasing AKI score was associated with worse outcomes, including increasing mortality, longer lengths of stay, and AKI non-recovery by 28 days. Taken together, these data highlight the importance of including urine output criteria in a comprehensive diagnosis of AKI. Furthermore, the presence of oliguria, even in isolation, seems to indicate a separate and possibly more severe phenotype of AKI than that without oliguria. In this way, oliguria can be used as a prognostic biomarker, identifying patients at higher risk for poor outcomes who may be appropriate for clinical trial enrollment or earlier intervention in therapy, such as renal replacement therapy. Unfortunately, despite these data, we do a terrible job at accurate urine output monitoring, even in the intensive care unit. This is highlighted by the results of this large study of over 15,000 patients that assess for the incidence and outcomes associated with intensive urine output monitoring, which they described as hourly urine output recording with no gaps for more than three hours in the first two days of ICU admission. The authors of this study found that this occurred in only 26 percent of patients, but when it did occur properly, patients were more likely to have appropriately identified AKI, and after adjustment for age and illness severity, intensive monitoring of urine output was associated with improved survival in patients with AKI and less fluid overload in all patients. To summarize, oliguria is a biomarker that should be leveraged to accurately identify AKI when present, enhancing AKI epidemiology. The presence of oliguria, both in isolation and in combination with serum creatinine elevation, pretends worse AKI and ICU-related outcomes. Thus, if we use both serum creatinine and urine output criteria, we can identify unique phenotypes of AKI with different prognostic implications, which have the potential to alter management, including through the identification of the highest risk phenotype, who may be appropriate for earlier consideration of more invasive therapies like renal replacement therapy. Finally, more intensive monitoring of urine output appears to improve AKI and ICU-related outcomes, and thus we should prioritize this important aspect of care for our patients. Building upon oliguria as a biomarker, back in 2013 Chawla and colleagues hypothesized that a patient's response to furosemide may be a cheap and easy way to assess tubular function early on in patients with AKI. This is because it is not effectively filtered at the glomerulus and needs to be actively secreted into the tubular lumen, presumably by a functional tubule, in order to be active. The goal was primarily to identify patients with tubular dysfunction early, such that this information could be used to enrich studies trying to answer the question of optimal renal replacement therapy timing. To do this, they gave one milligram per kilo of furosemide to loop diuretic-naive patients and one and a half milligrams per kilo to those who had received a loop diuretic in the past week, and assessed hourly urine flow rates for six hours afterwards, as well as progression to akin stage 3 AKI. You can see here that hourly urine output was significantly lower in patients who progressed to stage 3 AKI at each time point. The authors found that total urine output in the first two hours following furosemide predicted stage 3 AKI with an AUC of 0.87 and an optimal cutoff of 200 mls of urine during this time to predict non-responsiveness with impressive specificity and sensitivity as outlined here. In this same cohort of patients, the FST alone was shown to outperform a variety of biomarkers for stage 3 AKI prediction. Additionally, the performance of the FST plus biomarkers was not significantly better than just the FST alone. Importantly, the performance of the FST in critically ill adults has held up over time, as demonstrated in several studies. Two are outlined here. First, in this prospective study in 2019 looking at 92 patients with early AKI, they found similar test characteristics for the FST to predict progression to stage 3 AKI as the initial studies. Additionally, a meta-analysis has been performed of adult FST studies, and you can see here that the pooled sensitivity, specificity, and summary ROCs are quite impressive and superior to any single novel biomarker performance that is out there currently. Unfortunately, no prospective studies of the standardized FST have been performed in pediatric patients to date. However, a few retrospective analyses have shown similar findings to adults. In this multi-center retrospective analysis of children receiving furosemide after cardiac surgery by Penck and colleagues, urine flow rates were examined for six hours after indexed furosemide dose. The study authors found that two- and six-hour urine flow rates were lower in children who developed AKI versus those who did not, and those urine flow rates were also independently associated with the development of AKI after adjustment for multiple covariates. Specifically, the risk of AKI decreased by 41 percent with every 1 ml per kilo per hour increase in urine flow rate following furosemide. Thus, while there appears to be a relationship, more prospective work needs to be done to validate these findings in heterogeneous populations of critically ill children. So to summarize FST as a biomarker, the FST accurately predicts AKI progression in critically ill adults and is better than any available novel biomarker we have. Lower urine flow rates after furosemide in children following cardiac surgery are predictive of AKI development but needs prospective assessment in heterogeneous cohorts. And so if we build upon our previous use of urine output in serum creatinine to identify unique AKI phenotypes and then perform an FST in those with AKI, we can further risk stratify patients to identify the highest risk phenotypes, who again would be most appropriate for aggressive supportive therapy and future clinical trial enrollment. You will hear all about fluid balance, fluid overload, and its associated consequences today from Dr. Connors, so I will just touch briefly on the concept of fluid overload as a biomarker of AKI. If you'll crawl in Dr. Basu's AKI biomarker composite, percent fluid overload can be used as a biomarker of the body's system to continue to compensate for any ongoing injury. In other words, can the patient with AKI continue to main homeostasis via urine output and maintenance of a reasonable fluid balance? While Dr. Connor will outline the associations between fluid overload and AKI, recent work has sought to disentangle the independent effects of each, making its utility as a biomarker more compelling. Recently, Kat Gist and colleagues separated a cohort of 149 critically ill children into four AKI by serum creatinine and fluid overload subphenotypes at day three. As you can see depicted here, the addition of fluid overload, defined as greater than 20% fluid accumulation on day three, identified a subset of patients with longer lengths of stay and higher mortality compared to other AKI phenotypes. Furthermore, recognizing the importance of fluid balance on serum creatinine measurement, increasing numbers of groups are also starting to look at correcting serum creatinine for fluid status and noting increased strengths of associations between subsequent AKI and outcomes. Dr. Kathleen Liu and colleagues performed a post-hoc analysis of the FACT trial in 2011, which demonstrated that patients who were identified as having AKI only after adjustment for fluid balance had significantly higher rates of mortality and fewer ventilator free days compared to those without, as demonstrated in the red box. Importantly, these associations held up a multivariate analysis. Similarly, in the same cohort from the previous slide, Gist and colleagues used similar methodology to demonstrate that 21 patients class switched between AKI stages after serum creatinine was corrected for net fluid balance and corrected AKI classification demonstrated stronger associations between day three AKI and mechanical ventilation duration and ICU length of stay. Taken together, it appears that correction of serum creatinine for fluid balance or fluid overload may be important to enhance AKI detection and accurate sub-phenotyping of patients. In summary, when looking at fluid overload as a biomarker of AKI, it is important to note that it may be capable of identifying high-risk sub-phenotypes of AKI. Consideration should also be given to using fluid corrected serum creatinine to ensure prompt and precise detection of AKI. Again, if we return to this two-by-two figure with urine output and serum creatinine, we can instead use fluid adjusted creatinine. The risk stratification can also be further refined by the presence or absence of fluid overload as indicated by the gradients. Similar to previous slides, this phenotyping strategy can be used to identify the patients at highest risk for poor outcomes and direct our care and resources appropriately. Finally, moving to the most commonly used biomarker for AKI, serum creatinine, emerging data highlights the importance of considering the kinetics of serum creatinine as opposed to just static single values. Recently, Sutherland and colleagues looked at operationalizing the kinetic eGFR, whose formula requires two consecutive serum creatinine values as shown here. In this retrospective study of patients less than 21 years of age who underwent heart transplant over a 10-year period at Stanford, two kinetic GFRs were calculated, the first using a pre-op and the first post-op creatinine values, and the second using the first two post-operative creatinine values. While the performance of the kinetic GFR at both time points for AKI prediction was modest, with an AUC around 0.72 for both, what is interesting is shown in the figure on the right. You can see that the kinetic GFR falls more rapidly than the estimated GFR and reaches an eight around post-op day two, and then slowly starts to recover. In other words, demonstrating improved renal expiratory function. Though serum creatinine continues to rise at this time, albeit more slowly. This highlights the type of information that can be offered by looking at serum creatinine in a dynamic fashion as opposed to static. And finally, to round out our serum creatinine-based tools, I would like to just briefly touch on duration and severity, two tools that can be used to identify unique subsets of patients with prognostic implications. Recent ADKI consensus recommendations suggest that transient AKI, or in other words, AKI lasting less than 48 hours, be separated from persistent AKI, given literature to suggest differences in clinical importance and outcomes. Severity also needs to be considered as differential outcomes have been demonstrated when comparing mild versus severe AKI. However, the use of these tools in concert to elucidate unique AKI phenotypes has not been performed until recently. Basu and colleagues performed a secondary analysis of 750 patients from the AWARE study with sepsis and separated them into severity, mild versus severe, and duration, greater than or less than 48 hours, subphenotypes as shown here in these graphs on the left. As you can see here, 28-day mortality by severity and duration phenotypes demonstrated significant differences between the mild transient subphenotype and light blue and all other subphenotypes. Additionally, compared to mild transient, mild persistent had more mechanical ventilation, fewer ICU free days, and more complex ICU courses. And compared to severe transient, patients with severe persistent AKI had fewer ICU free days and more complex ICU courses. It is apparent from this work that severity and duration and potentially the interaction between them is important and accurate AKI prognostication, although further work is warranted. So how do we apply this at the bedside of a critically ill patient? We can first start at the time of ICU admission by asking the question, does the patient have serum creatinine-defined AKI? If yes, we can stage AKI by serum creatinine severity into mild or severe and can consider different escalations of therapy based on that. If the patient does not have AKI, we can consider whether or not they've received large volume resuscitation and consider adjusting for fluid balance. If they continue not to have AKI even with that, then we continue with urine output and serum creatinine monitoring and continue to reassess. As the patient is in the ICU for longer and we get more information, we can assess whether or not the patient is oliguric and restage AKI for risk stratification using that information. Additionally, we can consider an FST for risk refinement and as described in previous slides, categorize patients into one of eight AKI phenotypes that have differing risk profiles. And finally, once we hit 48 to 72 hours post-admission, if the patient still has AKI, it is now defined as persistent, which in and of itself re-risk stratifies the patient into a higher risk category. Additionally, we can assess whether or not the patient is fluid overloaded and consider risk stratification in the way described previously in this talk, identifying a higher risk subset of patients who are fluid overloaded and meet criteria for AKI with both serum creatinine and urine output criteria. I'd like to thank my mentors and the team from the Cincinnati Children's Center for Acute Care Nephrology, without whom all the work that I do would not be possible. I'm happy to take any questions. Thank you.

acute kidney injury

urine output

furosemide stress test

fluid overload

biomarkers

prognostication