Hello everyone. Thank you so much again for the introduction and to the organizers of this event for giving me the opportunity to speak today. These are my credentials, and again, I have no significant disclosures to report. In this talk, I'll be covering the topic of biomarker-based enrichment to target AKI, including novel biomarkers that I recognize are not yet widely available at many institutions. However, given the scope of this talk is to look at the future of precision AKI management, their inclusion is important as we examine what tools are on the horizon to improve the care of patients with and at risk for AKI. With this framework in mind, our learning objectives for this talk are to review the concepts of prognostic and predictive enrichment and how they may be used to employ a precision medicine approach to patient management, and to review novel biomarkers and tools that can be used within this precision medicine framework to individualize care of the patient with acute kidney injury. It is important to start with a few definitions before we delve into the literature in this area, first being precision medicine, which refers to the prevention, diagnostic, and treatment strategies that take individual patient characteristics into account. Key to precision medicine is the concept of enrichment, which is selection of a subgroup of patients more likely to respond to a given therapy or suffer an outcome of interest when compared to an unselected population. Enrichment can further be broken down into prognostic enrichment, which is the selection of patients with a higher likelihood of having a disease or disease-related outcome of interest, for example, patients who are more likely to develop persistent AKI, or predictive enrichment, which is the selection of patients who are more likely to respond to a given therapy on the basis of biological mechanism, and a very common example of this is the use of trastuzumab for HER2-positive breast cancer. So how do we operationalize these concepts of precision medicine and enrichment at the bedside of critically ill patients? This is a figure from our recent review outlining a precision medicine approach to biomarkers and substance-associated AKI that can be applied to AKI broadly as a whole. You start with a heterogeneous cohort of patients with AKI and then apply a prognostic enrichment tool, which separates patients into high-risk or low-risk for your outcome of interest. In patients who are deemed to be at high risk, you can further apply predictive enrichment tools, i.e., those that identify specific biology of patients, in order to apply a specific treatment based on patient biology. We'll start by discussing prognostic enrichment tools for AKI that can readily be applied at the bedside now. These include the Renal Angina Index to guide tubular injury biomarkers, TIM-2 IGF-BP7 for identification of early kidney stress, and the novel biomarker CCL-14 for prediction of severe persistent AKI. As we cover the different prognostic enrichment tools for AKI, I ask you to consider these patients, and we'll bring them up throughout the discussion. First is patient 1, which is a 10-year-old female who is a day plus 40 from bone marrow transplant for relapsed AML, who presents to the pediatric ICU as a transfer from the floor with septic shock. She has received three 20-mL per kilo fluid boluses and is on an epinephrine infusion at 0.05 mics per kilo per minute. She has a serum creatinine of 2 milligrams per deciliter, which is more than two times her baseline, and is making some urine. Patient number 2 is a 50-year-old male with history of diabetes and hypertension who is admitted with COVID pneumonia requiring invasive mechanical ventilation. His serum creatinine is 1.2 on admission with a baseline of 0.9. He is on broad-spectrum antibiotics with vancomycin and piptazo. And then finally, patient 3 is a previously healthy 60-year-old male who presents to the ICU as a transfer from the surgical floor following abdominal surgery two days ago with concerns for septic shock. On arrival to the ICU, it is noted that his urine output has been downtrending, and his serum creatinine is 3 milligrams per deciliter with a baseline of around 1. He has required 3 liters of fluid resuscitation and has been initiated on a norepinephrine infusion to maintain his blood pressure. I wanted to start with the renal angina index, as it is a prognostic enrichment tool for AKI that can be applied even in low-resource settings and in a variety of patients. Over a decade ago, Stu Goldstein and Ming Chala came up with the concept of renal angina in an attempt to risk-stratify patients within a similar framework to the myocardial infarction paradigm. That is, can we combine clinical signs or kidney function changes to identify patients with ongoing renal angina, which unlike patients with an MI, does not hurt, and then use that information to inform novel biomarker measurement, similar to how chest pain informs troponin measurement. To achieve this goal, they derived and validated the renal angina index in several cohorts of critically ill children. Estimated at 12 hours after ICU admission as the product of a risk score and an injury score, the renal angina index predicts severe AKI 72 hours later, as outlined in subsequent slides. Data from these initial derivation and validation cohorts demonstrate that the predictive capacity of the renal angina index for severe AKI at day 3 was pretty good, with AUCs around 0.8. Furthermore, the negative predictive value of this tool was excellent and near to 100% in some cohorts, suggesting it is a great tool to rule out those patients who are unlikely to develop severe AKI at day 3. More recently, this tool was prospectively validated in a cohort of nearly 1,000 patients from the multinational AWARE study. The predictive performance of the RAI was again similar, with an AUC of 0.83, and when compared to context-free serum creatinine elevation alone at the same time point, screening positive for renal angina was associated with a higher risk of developing severe AKI three days later. Getting back to the initial hopes for the renal angina index construct, how does it perform when used to guide injury biomarker assessment? A single-center, prospective study of nearly 200 critically ill children assessed the predictive performance of the RAI with various biomarkers. In this study, the inclusion of NGAL at an optimal cutoff of 150 ng per mL doubled the positive likelihood ratio of the RAI alone and improved the AUC to a staggering 0.97 for severe AKI predictions 72 hours later. Given the increasing availability of NGAL and the relative ease with which the RAI can be calculated, this is a feasible stepwise prognostic enrichment tool that can be operationalized at the bedside of patients now. At our center, Dr. Stu Goldstein and his team have taken these data and done just that, applied it to the bedside of patients in our PICU with the Taking Focus II study. As you can see in this schematic, patients have a renal angina index assessed at 12 hours after ICU admission, and if it is low, they continue with standard critically ill patient management. If the RAI is elevated at 8 or higher, they undergo a urine NGAL assessment, and if it is low, they continue with standard management, but if it is elevated, the patients either undergo a furosemide stress test or begin early fluid restriction and consideration of renal replacement therapy when at 10 to 15% fluid overloaded. The patients also undergo daily urine sample collection if their initial NGAL is elevated for renal recovery markers. It is important to note that the RAI and reflux NGAL have been automatically operationalized into the electronic medical record workflow, requiring little effort on the part of providers and bedside staff. When we prospectively assess the Taking Focus II paradigm performance using 1,600 patients from July of 2018 to June of 2019, we found that being RA positive and NGAL positive was associated with significantly higher rates of severe AKI at day three and the need for renal replacement therapy compared to all other patients in the cohort, which includes RAI negative and RAI positive patients who did not have an elevated NGAL. Furthermore, different from previous studies, the RAI positive NGAL positive had an impressive specificity at 98%, as well as the other test characteristics outlined here, and these data are unpublished and presented with permission of Dr. Goldstein. So the tool performs well, but what has it done for our patients? When we assessed patient outcomes for those requiring CRT in the pre- and post-TFII period, we found that we were starting CRT earlier at a median of 1.6 days from ICU admission compared to a median of three days from ICU admission in the pre-Taking Focus II period. However, despite this, our total CRT duration was actually lower at a median of 3.7 days compared to a median of 5.5 days prior to this project. ICU length of stay was also lower for patients requiring CRT in the post-TFII period. And finally, trends in the data also suggest improvement in survival to both CRT end and ICU discharge in the post-TFII period compared to prior to study initiation. And again, these are unpublished data presented with the permission of Dr. Goldstein. So how do we apply these constructs to patient care? Going back to patient one, this was the 10-year-old female who was status post-obomer transplant with septic shock, who had gotten a lot of fluid resuscitation, was on an epinephrine infusion, and had severe AKI bicreatinine. If we go back to the RAI, we can score this patient as a risk level of 5 and an injury level of 8, which gives us a maximum score of 40 for the renal angina index. Therefore, we can identify this patient as being high risk for persistence of severe AKI in 72 hours. We can select this patient to send a tubular injury biomarker for further risk stratification, such as NGAL, and consider a furosemide stress test to guide consideration of earlier CRT for fluid restriction in non-responders. And we can implement standardized kidney protection as much as possible. In the interest of time, I will also just note that modifications to the renal angina index have been examined in special populations, namely in critically ill adults, pediatric septic shock, and more recently in children post-cardiopulmonary bypass. As the risk factors for AKI are going to be different across different populations, reworking of this risk stratification tool is likely needed for other unique populations, for example, neonates. And this work is ongoing. Moving on from the renal angina index, another prognostic enrichment tool that can be used before severe AKI development is TMP2-IGFBP7, which are markers of kidney stress. Back in 2013, Kishani and colleagues published this paper out of the SAFIRE study that validated the performance of combining the cell cycle arrest markers TMP2 and IGFBP7 to predict severe AKI with an overall AUC of 0.8, similar to that of the renal angina index. Similar to the RAI NGAL framework, the addition of the TMP2-IGFBP7 tool to a clinical risk model significantly improved AKI prediction. It is important to note, however, that this tool was used only in patients who did not have stage 2 to 3 AKI as part of this study. These results have subsequently been validated in additional cohorts, which has led to this tool being the first FDA-approved AKI risk stratification tool for adults known as NefraCheck. A recent paper sought to look at the clinical use patterns of this tool in AKI management in both North America and Europe. Based on expert consensus and survey, the following indications for measurement and potential standardized nephroprotective bundle use in response to increased risk for severe AKI are shown. As you can see, the authors suggest performing a NefraCheck evaluation in patients who are critically ill with shock, sepsis, are postoperative from major or cardiac surgery, those who have had trauma, and those with other cardiovascular respiratory compromise. If the patients are noted to be moderate or severe risk for AKI, they recommend the following components of a nephroprotective bundle and continuing to assess AKI risk increase and intervene when possible. Patients who are deemed low risk can continue a standard of care and repeat assessment with NefraCheck test if additional exposures occur. It is important to note, though it's not included in this figure, that this tool has been used as a prognostic enrichment tool for clinical trial enrollment with success in some populations. You may have noticed that the nephroprotective bundle on the previous slide included what feels like several common sense interventions in a critically ill patient at risk for AKI, such as avoiding nephrotoxins, maintaining adequate hemodynamics, and consulting nephrology. However, numerous studies have now demonstrated that we do a poor job at performing these common sense interventions, and these two studies in particular demonstrate that the application of standardized bundles in high risk patients can actually improve outcomes and reduce AKI burden. In these two randomized controlled trials of post-surgical patients, patients without AKI underwent TMP2-IGF-BP7 screening, and if found to be high risk for AKI, were randomized to either standard of care or standardized kidney protection bundle, as outlined on the right. As you can see, this bundle was fairly simple. However, in both studies, the intervention group had significantly lower incidence of AKI, including severe AKI. These studies provide evidence that being intentional about common sense nephroprotection has the potential to improve patient care. So how do we apply this to some of our patient cases? If you recall, patient two is a 50-year-old with some past medical history and comorbidities who has COVID pneumonia and is intubated requiring mechanical ventilation. He does not have AKI, but does have a slight increase in his serum creatinine and is on some nephrotoxic agents. First, you can identify this patient as appropriate for TMP2-IGF-BP7 screening to assess for kidney stress. If elevated, we can implement standardized nephroprotection bundle and evaluate for and mitigate ongoing kidney stressors. And if available at your institution, could consider utilizing this to enrich enrollment in clinical trials in the future. Since TMP2-IGF-BP7 has really only been validated for AKI prediction in those without serum creatinine-defined AKI, it is important to have prognostic enrichment tools to determine which patients with AKI will have AKI persistence. CCL14 was identified and validated as part of the Ruby and Sapphire studies and plays a role in macrophage trafficking, potentially contributing to maladaptive repair and persistent kidney damage. As such, it has been demonstrated to predict persistence of severe AKI in those with early severe AKI with a similar AUC to the other tools already presented around 0.8. Furthermore, not only did CCL14 predict persistence of AKI, separation of patients into tertiles by CCL14 concentrations also demonstrated increased risk of mortality in renal replacement therapy with increasing concentrations. Using these data and the biological activity of CCL14, it is also possible that this tool is a prognostic enrichment target for need for renal replacement therapy and thus could enrich future clinical trials looking at optimal RRT timing in the development of MAKE90 outcomes, including AKD and CKD. It's also possible as we learn more about CCL14 that it could be a predictive enrichment target and something that could be modified along the pathway of AKI. So how do we apply this to a patient? If you recall, patient three was a previously healthy 60-year-old who's on the surgical floor and presents the ICU with sepsis. On arrival to the ICU, he has serum creatinine-defined AKI and is on norepinephrine to maintain his blood pressures. We've identified this patient as appropriate for CCL14 screening to evaluate risk for persistence of his severe AKI. If it is elevated, we can consider earlier renal replacement therapy since we know he's at higher risk for continuing to have severe AKI in three days and can also implement standardized nephroprotection, provide appropriate counseling to the patient and the family regarding trajectory. And finally, as mentioned on the last slide, we can consider utilizing this tool to enrich enrollment into clinical trials, especially those aimed at renal replacement therapy timing, a topic that has been controversial and has had several studies showing different results. While prognostic enrichment is important to improve the identification of patients appropriate for aggressive kidney protection, earlier escalation of therapy, long-term follow-up, and informed enrollment in future therapeutic trials, what we really need to identify are predictive enrichment strategies. Those that identify the underlying biological mechanisms such that they may be targeted by specific therapies. However, these are currently very limited to nonexistent. An opportunity for potential predictive enrichment comes from the use of tubular injury biomarkers. Expert consensus out of ADK23 resulted in proposed changes to the diagnostic criteria for AKI that incorporates both functional markers like serum creatinine and urine output, as well as damaged biomarkers such as NGAL. There has been a fair amount of work in both adults and children that have demonstrated differences in prognostic implications for these four damage and functional biomarker-based AKI phenotypes, which may also be able to guide therapeutic decision-making. So how do we use these biomarkers in practice? As outlined on the previous slide, functional and injury biomarkers can be used to characterize patients into one of four AKI phenotypes as outlined here, each of which can inform care. For example, patients with functional AKI may be a subset of patients who are appropriate for more aggressive fluid resuscitation without concern for significant risk of fluid overload, whereas a similar patient with evidence of damage-associated AKI may be someone in whom you consider earlier pressors over more fluid resuscitation, fully placement for strict eyes and nose management, and judicious use of nephrotoxins with monitoring of levels if possible. Patients with subclinical AKI or those with damage biomarker elevation in the absence of serum creatinine rise are more challenging, as little evidence exists to guide who to assess damage biomarkers in when functional changes are not present. However, data suggests that outcomes for this group are more similar to the damage AKI as opposed to the functional AKI group, and thus they represent a subset of patients who should be managed similarly even in the absence of serum creatinine rise. I'd like to end by touching on an exciting on-the-horizon example of predictive enrichment for patients with vasodilatory shock, especially those with AKI. Angiotensin-2 is a novel vasoactive agent that has been demonstrated to improve blood pressure in adults with catecholamine-resistant vasodilatory shock. Belomo and colleagues performed a post-hoc analysis of the ATHOS-3 trial examining efficacy of this drug to assess the hypothesis that renin levels as a surrogate of low angiotensin-2 could serve as a predictive enrichment tool to identify patients most likely to respond. This hypothesis is grounded in the fact that endothelial injury occurs in the setting of shock, which results in angiotensin-converting enzyme dysfunction, decreased conversion of angiotensin-1 to angiotensin-2, and subsequent upregulation of renin. And indeed, when the authors tested their hypothesis, they found that patients with initial renin concentrations above the study population median had uniformly improved outcomes, including liberation from renal replacement therapy if they received angiotensin-2 compared to placebo. And furthermore, when they looked specifically at a subset of patients with AKI requiring renal replacement therapy at the time of drug initiation, they found improved 28-day survival, increased rates of CRT liberation by day 7, improved mean arterial pressure and decreased vasoactive need in the first 3 hours, more freedom from vasoactives at day 7, and more freedom from mechanical ventilation at day 7 in those who received angiotensin-2 compared to placebo. There is some physiologic basis for these associations, as angiotensin-2 is known to have direct effects on intrarenal circulation, with preferential vasoconstriction of the efferent arterial, and increased in glomerular perfusion pressure, which may help restore GFR in vasodilatory shock states such as sepsis. Putting this all together, it is possible that earlier renin-guided administration of angiotensin-2 in patients with vasodilatory shock and AKI may improve AKI and ICU-related outcomes. This would be a clear example of predictive enrichment for patients with AKI that is sorely needed to improve care. To summarize, validated prognostic enrichment tools for AKI in a variety of settings exist and should be utilized routinely for risk assessment in standardized management in patients with or at risk for AKI. These tools should also be considered for use in identification of patients appropriate for therapeutic clinical trial enrollment targeting AKI. And finally, predictive enrichment tools for AKI management are sorely needed, and when possible, we should leverage existing prognostic enrichment tools to help identify and assess them. I'd like to thank my mentors and the team from the Cincinnati Children's Center for Acute Care Nephrology, without whom all of the work that I do would not be possible. I'm happy to take any questions. Thank you.

biomarker-based enrichment

acute kidney injury

precision medicine

prognostic enrichment

predictive enrichment

CCL14 biomarker