Proceedings of the 2022 UAB CRRT Academy: Non-Invasive Hemodynamic Monitoring to Guide Fluid Removal with CRRT and Proliferation of Extracorporeal Blood Purification Devices

The University of Alabama at Birmingham (UAB) Continuous Renal Replacement Therapy (CRRT) Academy was founded by Dr. Ashita Tolwani in 2007. Since then, every autumn, a 2-day conference has welcomed nephrology and critical-care fellows, practicing clinicians, nurses, pharmacists, and other healthcare professionals from across the USA and the world, growing to nearly 140 attendees from 8 countries in 2022. Along with lectures, the conference consists of hands-on workshops, simulations, and interactive problem-solving sessions to provide an in-depth understanding of CRRT. The goals of this conference are twofold: (1) to teach the fundamentals of CRRT to help trainees rapidly develop practical skills they can use immediately in the care of patients and (2) to enhance the knowledge and skills of practitioners with preexisting competence in CRRT to help them become local experts and manage their CRRT programs.

As we celebrate the 15th anniversary of the academy, we reflect on the evolution of the conference. The COVID-19 pandemic led to restructuring over the past 2 years, resulting in a hybrid format in which faculty traveled to Birmingham while most of the conference attendees participated virtually. This format preserved the interactive nature of the conference while mitigating risk for participants. Though the pandemic was a major educational and logistical challenge, the virtual platform allowed for increased numbers of participants from more locations. Prior to COVID-19, the in-person conference was limited to 50–75 attendees per year, including 1–2 international participants. During the pandemic, to allow for full COVID-19 precautions (including distance seating and individual box lunches), in-person participants were limited to approximately 20 local attendees. However, using the virtual platform, nearly 120 additional participants were able to attend, including attendees from South America, the Middle East, and Asia. To adapt to the challenges of remote learning, we introduced a series of educational innovations during the pandemic, including a faculty-operated interactive “chat box.” While the faculty members were all present in person at UAB, this chat box allowed remote attendees to maintain dynamic interactions and obtain real-time responses to questions. Additionally, we introduced a “Shark Tank” contest in which multiple entrants from around the world pitched CRRT-related research ideas to a panel of faculty judges.

This year’s most significant innovation was the introduction of a keynote speaker which provided the attendees with an introduction to cutting-edge research driving the evolution of CRRT practice, while also underscoring the need for CRRT providers to engage in local quality-assurance activities including monitoring of their local CRRT program data and outcomes. We welcomed Dr. William Beaubien-Souligny from Montreal as our first keynote speaker to present “Fluid management during CRRT: The value of quality care and non-invasive hemodynamic monitoring.” Additional emerging concepts that arose in this year’s conference include (1) the growth of extracorporeal blood purification (EBP) devices to potentially expand the indications for CRRT beyond acute kidney injury (AKI); (2) improving the efficacy and applicability of regional citrate anticoagulation (RCA), including safely adapting RCA to patients with liver disease; and (3) harnessing hybrid modalities of renal replacement therapy (RRT) (i.e., prolonged intermittent RRT, or PIRRT) to optimize de-escalation from CRRT in patients recovering from critical illness.

To broadly disseminate a slice of the CRRT Academy experience and provide a state-of-the-art review of select emerging topics in the field of critical care nephrology, we introduce the proceedings of the UAB CRRT Academy. Our focus on this first iteration will be on two concepts: (1) non-invasive monitoring to guide fluid removal by RRT and (2) novel EBP devices.

A large and growing number of observational studies have established that positive fluid balance and volume overload in patients with AKI [1‒12], acute respiratory distress syndrome (ARDS) [13‒15], sepsis [11, 15‒22], and general critical illness [23‒25] are associated with poor outcomes, including progressive kidney dysfunction, lower odds of kidney recovery, prolonged mechanical ventilation, and higher mortality. In all these analyses, fluid overload remains an independent predictor of mortality or other harm after adjustment for other covariates including severity of acute illness. The possible negative effect of fluid overload on kidney function has been attributed to venous congestion leading to renal venous hypertension, increased renal interstitial pressure, and ultimately reduced renal perfusion pressure, renal blood flow, and glomerular filtration rate, as suggested by multiple studies associating elevated central venous pressure (CVP) in critically ill patients with worsening renal function [4, 16, 26‒32]. Likewise, it has been proposed that virtually all organs may be adversely impacted by fluid overload and venous hypertension in critical illness (Fig. 1) [33].

Limited prospective data demonstrate that fluid restriction is safe, if not universally beneficial, in critical illness. The Fluid and Catheter Treatment Trial (FACTT) showed that fluid restriction as tolerated in patients with ARDS resulted in improved oxygenation and shortened duration of mechanical ventilation by 2 days with a trend (p = 0.06) towards less need for RRT [34]. More recently, the multicenter CLASSIC trial randomized over 1,500 patients with septic shock, after initial resuscitation, to restriction of resuscitative fluid compared to usual care and found no difference in mortality, rates of AKI, or serious adverse events [35]. Furthermore, in the multicenter REVERSE-AKI pilot trial, the proactive use of diuretics and limitation of fluid intake to maintain neutral daily fluid balance in the setting of established AKI resulted in decreased need for RRT (relative risk 0.42, p = 0.043) [36].

Nonetheless, guidelines, based largely on observational data, continue to emphasize the need for early resuscitation in septic shock [20, 21, 37‒47]. This has led to the conceptualization of temporal phases of fluid therapy in sepsis or critical illness (Fig. 2a) [48‒51]. While large randomized controlled trials (RCTs) have been recently completed [35, 52] or are ongoing (NCT04569942, NCT05179499) to determine the optimal approach to the early resuscitation or optimization phases, relatively few interventional data exist to guide the optimal approach to the de-resuscitation phase, especially in patients with AKI requiring RRT [53, 54]. However, two pilot RCTs are currently active. The RELIEVE-AKI trial (NCT05306964) will compare two strategies of fluid removal or net ultrafiltration (UF) with CRRT, while the probe-fluid trial (NCT05473143) will compare a protocolized fluid removal strategy to usual care.

The use of RRT to achieve de-resuscitation may benefit patients by limiting the severity or duration of fluid overload, but the potential benefits of treating volume overload must be balanced against the potential risks of excessive fluid removal which can potentially lead to intradialytic hypotension and end-organ hypoperfusion (Fig. 1, 2). For example, myocardial stunning – a phenomenon that is well described in patients with end-stage kidney disease (ESKD) undergoing intermittent hemodialysis (IHD) and associated with higher UF rates, intradialytic hypotension, myocardial hypoperfusion, and higher mortality [55‒57] – has recently been demonstrated in patients with AKI undergoing IHD and CRRT [58, 59]. Indeed, while the ability to achieve negative fluid balance with CRRT is associated with higher survival and failure to achieve the prescribed fluid balance goal is associated with lower survival, the relationship between the rate of fluid removal with CRRT and outcomes is complex, with some studies demonstrating that higher CRRT fluid removal rates are associated with mortality and others demonstrating that lower CRRT fluid removal rates are associated with worse outcomes [7, 60‒65].

Adding to the complexity of this delicate balance, the pathophysiology of hypotension during RRT is complex, with only a subset of the hypotensive episodes that occur during RRT being due to preload dependence [66, 67], suggesting that some patients may benefit from ongoing volume removal with RRT despite hypotension and underscoring the need for accurate individualized measures of volume status (Fig. 3). However, traditional measurements of volume status, such as a physical examination or static measures such as CVP, tend to perform poorly at predicting response to intravenous fluids in critically ill patients [68, 69]. Though studies assessing tolerance of fluid removal are limited, traditional static measures of volume status may similarly fail to predict dynamic tolerance of fluid removal with RRT, contributing to substantial worldwide practice variation in fluid management with RRT [70‒72]. While the focus of this review is on the tools to guide volume removal with RRT in the de-resuscitation phase of critical illness, interested readers are directed towards a series of recent high-quality reviews dealing with important background elements, including the general approach to fluid therapy in AKI, the optimal approach to earlier phases of fluid therapy in the ICU, choice of fluids in critical illness, and the use of diuretics in de-resuscitation [27, 73‒76].

Recently, point-of-care ultrasound (POCUS) has been proposed to be a useful adjunct in assessing volume status and guiding volume removal by RRT, both to detect patients likely to tolerate net UF and patients with congestion in need of urgent fluid removal [77, 78]. POCUS can be used in four ways to assess volume status, including lung ultrasound (LUS), inferior vena cava (IVC) ultrasound, venous Doppler, and echocardiographic estimation of stroke volume (Table 1). These tools are used either to detect venous congestion or assess fluid responsiveness (FR). FR is defined as an increase (usually by ≥ 10–15%) in stroke volume or cardiac output in response to fluid administration. Notably, FR is normal and is not an indication for intravenous fluids per se, but it serves to identify patients with potential to benefit from fluids. Conversely, lack of FR largely excludes any potential benefit from fluids but may identify patients likely to tolerate fluid removal with RRT.

LUS can be used to detect “B-lines,” vertical line artifacts representing fluid in the interlobular septa of the lung parenchyma (Fig. 4a). Studies of LUS in patients with AKI have been limited, with two studies demonstrating that B-lines correlate with hypoxemia and fluid overload as determined by weight increments [81, 82]. LUS has been more extensively evaluated in patients with ESKD, in which B-lines have been shown to be more sensitive for lung congestion than auscultation, to be associated with adverse outcomes, and to decrease rapidly with UF during hemodialysis [107‒110]. Recent RCTs in maintenance HD have shown that UF guided by LUS is more effective than standard care at lowering dry weight and ambulatory blood pressure with decreased rates of intradialytic hypotension, but data demonstrating cardiovascular or mortality benefit thus far have proven elusive [79, 80]. However, in addition to limited data in AKI, the utility of LUS to guide UF in the ICU may also be limited by specificity. Though some ultrasound findings may help differentiate hydrostatic pulmonary edema, diffused B-lines can also be seen in other conditions common in the ICU such as ARDS or viral pneumonia [111‒115].

Ultrasound of the intrahepatic IVC, which, like LUS, requires relatively minimal skill, can be used to estimate right atrial pressure (RAP) in spontaneously breathing patients, with IVC diameter >2.1 cm and <50% collapse with respiration consistent with high RAP (10–20 mm Hg) and IVC diameter ≤2.1 cm and >50% collapsibility consistent with low RAP (0–5 mm Hg) [116]. However, such estimations of RAP are crude, and the use of IVC ultrasound to determine FR has significant limitations, including being unreliable in a variety of settings such as intra-abdominal hypertension or spontaneous breathing with unpredictable changes in intrathoracic pressure [117‒121]. Nevertheless, the use of IVC collapsibility index (IVC-CI) to guide fluid removal has been studied somewhat in both ESKD and AKI patients [122, 123]. In a pilot study of 24 patients undergoing slow continuous UF for acute decompensated heart failure with oliguria and diuretic resistance, IVC-CI >30% predicted hypotension with an estimated 100% sensitivity and 95% specificity, though the precision of these estimates was limited [123].

Given the limitations of isolated IVC assessment, it has been combined with venous Doppler of the hepatic, portal, and intrarenal veins into the Venous Excess Ultrasound (VExUS) grading system (Fig. 4a) [78, 83]. VExUS detects the characteristic changes in pulsatility, direction, and/or duration of venous flow seen with increasing levels of congestion at these sites. Though prospective data demonstrating the utility of VExUS to guide volume management remain limited, many observational studies in the setting of heart failure, critical illness, or cardiac surgery have shown the VExUS components to correlate well with other hemodynamic parameters – such as CVP and renal perfusion pressure, NT-pro-brain natriuretic peptide levels, and echocardiographic measures of right ventricular function – and to be independently associated with clinical outcomes, including rates of diuretic resistance, AKI, major adverse kidney events (a composite of death, need for RRT, and persistent kidney dysfunction), cholestatic congestive hepatopathy, postoperative complications, rehospitalization, and death [83‒92, 124‒127]. Importantly, VExUS has multiple limitations, including a variety of conditions that must be considered during the interpretation of venous Doppler, including severe tricuspid regurgitation, severe pulmonary hypertension, chronic liver disease, chronic kidney disease, or obstructive nephropathy [128‒130]. Other limitations include that the portal vein can exhibit pulsatility in healthy adults and that intrarenal vein Doppler is often not technically feasible [130, 131]. Therefore, interpretation of VExUS should ideally include as many elements of the grading system as possible and require integration of all other available clinical data [77, 78].

Finally, POCUS can be used to estimate stroke volume and cardiac output by measurement of the velocity-time integral (VTI) through the left ventricular outflow track (LVOT), in which pulsed wave Doppler is used to estimate flow through the LVOT or aortic root (Fig. 4b). Tracking changes in LVOT VTI in response to fluid administration or a variety of other maneuvers can be used to assess FR [132]. Among the most commonly employed maneuvers utilized to manipulate preload to permit assessment of FR is the passive leg raise (PLR), in which both legs are raised using the bed to 45° to provide a reversible test bolus of approximately 300 mL of blood [133]. PLR-induced change in cardiac output, as estimated by a variety of methods including specifically LVOT VTI, has been shown in many studies to be among the most accurate methods for assessing FR in critically ill patients [69, 132, 134‒137]. However, though the use of PLR to guide fluid removal coupled with other methods to estimate stroke volume has been evaluated and serial VTI to guide fluid removal has been proposed as superior to standard hemodynamic monitoring, no prospective trials exist analyzing the utility of LVOT VTI to guide overall fluid management with RRT [99‒101, 138].

Like POCUS, bioimpedance and bioreactance are novel and completely non-invasive tools to assess volume status. In both cases, surface electrodes are applied to the chest wall, and high-frequency alternating current is passed through the thoracic cavity. Bioimpedance analysis (BIA) estimates stroke volume from the change in electrical impedance that results from the change in thoracic blood volume that occurs with each cardiac cycle [139, 140]. Bioreactance, a more sophisticated technology, analyzes the relative phase shift of the alternating current to estimate aortic blood flow and produces a signal that is less affected by confounding factors such as electrode placement, respiration, or other movement [140, 141]. Data to support the use of these devices for fluid management in the ICU are mixed [142‒145], but a recent multicenter RCT did demonstrate that a protocol of assessing FR with PLR and a bioreactance device can be used to safely restrict fluid administration in patients with septic shock [98].

BIA with electrodes applied to the limbs can also be used to estimate whole-body and extracellular water content in patients with AKI in the ICU, though available data in this context are limited. Several observational studies have redemonstrated using BIA that fluid overload at CRRT initiation or failure to achieve net UF with CRRT are associated with mortality [94‒96]. In an RCT of 65 patients, BIA to guide fluid removal with CRRT did produce lower body water at the end of CRRT, but the protocol had no detectable impact on clinical outcomes [97].

Relative blood volume monitoring uses devices applied to the RRT circuit that provide an online estimate of hematocrit which is extrapolated to generate a continuous estimate of relative blood volume. Though popular, these devices have yielded mixed results when used to guide fluid management in ESKD or AKI settings [102‒106, 146‒149].

Finally, though minimally invasive rather than non-invasive (i.e., requiring arterial and central venous catheters), pulse contour analysis calibrated by transpulmonary thermodilution (e.g., PiCCO, Getinge AB, Goteborg, Sweden) can provide accurate estimations of stroke volume and cardiac output. Though generally considered very accurate in determining FR, studies of the PiCCO device to predict tolerance of fluid removal with IHD, PIRRT, or CRRT in critically ill patients with AKI have yielded mixed results [99‒101, 140].

While accumulating epidemiological data suggests that fluid overload mediates adverse outcomes in critically ill patients, the optimal fluid management strategy often remains elusive in clinical practice. The prescription of fluid removal with RRT – be it IHD, PIRRT, or CRRT – is a complex medical intervention that involves multiple interacting components and dynamic assessments (Fig. 5). Though the ability to reliably assess volume status is a fundamental first step, the determination of fluid status and target will not improve patient outcomes unless it is accompanied by a fluid management strategy that can be feasibly incorporated into routine patient care and can be employed continually or serially to allow for iterative reassessment as critical illness evolves. Indeed, as volume status is often dynamic in critical illness, a potential pitfall of reliance on POCUS and other hemodynamic tools which provide a static snapshot of volume status is the failure to repeat assessment when indicated. On the other hand, though dynamic measures are clearly superior to static measures in predicting responsiveness to fluid administration [150], these dynamic measures have not yet been validated to guide fluid removal. This level of complexity may partially explain why fluid management remains one of the most challenging and persistent problems in clinical nephrology and critical care medicine. Future studies must not only focus on improving our ability to diagnose volume disturbances but also on the development, validation, and optimal implementation of fluid management strategies in patients requiring extracorporeal fluid removal with RRT.

EBP – the use of various devices to clear the blood of pathogens, inflammatory mediators, and toxins – gained increased traction during the COVID-19 pandemic. The basis for EBP comes from emerging knowledge that organ crosstalk in critical illness is responsible for significant end-organ damage and the resulting theory that halting this process may interrupt the inflammatory cascade of events that ultimately leads to immune dysregulation and multiorgan failure [151]. In the following sections, we delve into the pathophysiology of the inflammatory cascade and distant organ damage and describe some of the newer EBP devices and applications that have been developed and studied to mitigate the damage process. We focus primarily on the adsorptive and immunomodulatory EBP devices discussed during the CRRT Academy, and we acknowledge that, though broad, this list is not exhaustive of all devices or applications in the rapidly evolving and fascinating field of EBP.

When a host is infected, immune cells recognize specific patterns on pathogen surface membranes, patterns known as pathogen-associated membrane proteins (PAMPs) [151, 152]. Recognition of PAMPs represents the inception of the “cytokine storm.” The activated immune cells subsequently produce anti- and pro-inflammatory cytokines such as tumor necrosis factor-alpha, interleukin-1 (IL-1), IL-6, IL-8, and IL-10. Injured host cells also express damage-associated molecular patterns (DAMPs) which enhance leukocyte activation and cytokine production. This circular activation leads to immunoinflammatory dysregulation (Fig. 6) [151, 152].

Several novel technologies have been developed, each targeting specific stages of the inflammatory cascade (Table 2). These include filters which can directly remove pathogens, cytokines, and endotoxin and devices which aim to modulate the immune system. The filters employ varying principles for clearance which include diffusion, convection, and adsorption, but the focus of this discussion is on adsorptive technologies. Adsorption is a method for removal of molecules via attachment to various sorbents because of chemical affinity to ion-exchange resins and chemisorbents [153]. Sorbents and adsorptive technologies provide a means to remove highly protein-bound solutes which are not effectively cleared by diffusion or convection [154].

These principles of adsorptive EBP have been applied for decades in two realms, namely hemoperfusion (HP) for the treatment of intoxications and extracorporeal liver support devices. HP has an enhanced ability relative to HD to clear poisons which are lipophilic and/or highly protein-bound. Though HP continues to be used commonly in specific settings (e.g., paraquat poisoning in developing countries), HP has been used rarely in the USA since the 1990s, where HD is far more commonly used to treat toxic ingestions [155]. Liver support devices function by removing albumin-bound solutes. The most common is the molecular adsorbent recirculating system (MARS), which was developed in the 1990s and can be used for refractory cholestatic pruritus or drug intoxications, especially when accompanied by severe liver injury (e.g., acetaminophen or Amanita mushrooms) [156]. However, MARS is most commonly used for acute or acute-on-chronic liver failure, in which it can improve hepatic encephalopathy and serve as a bridge to liver transplantation. Detailed discussion of HP for intoxication or extracorporeal liver support devices is outside the scope of this review, but readers are directed to other recent reviews [155‒158].

More recently, EBP technologies have been applied broadly to critically ill patients with septic shock or multiorgan dysfunction. The development and study of these novel EBP devices accelerated during the COVID-19 pandemic. Filters are now available for removal of pathogen, endotoxin, and cytokines and for immunomodulation in addition to traditional organ support (Table 2).

The Seraph^® 100 Microbind^® Affinity Blood Filter is an extracorporeal HP device composed of 0.3-mm polyethylene beads bound to immobilized heparin [159]. This mimics the negatively charged heparin sulfate naturally found on mammalian cells, serving as a binding site for basic amino acids and surface proteins which are often found on bacteria, viruses, and proteins. The Seraph beads carry out a similar role as heparin sulfate and irreversibly bind pathogens passing through the filter, thereby removing them from the bloodstream [159]. Drug removal has also been evaluated, and while aminoglycosides may be adsorbed, a spectrum of 18 antifungal, antibacterial, and antiviral agents all had neglectable rates of removal from systemic circulation [160]. Though the heparin release produces insignificant anticoagulation, Seraph is contraindicated in heparin allergy [161]. Data to support the clinical use of the Seraph 100 filter are currently limited to observational studies. In one study of 53 patients with severe COVID-19, compared to 53 matched controls, the use of Seraph was associated with an increase in vasopressor-free days and mortality, though the differences were attenuated when adjusting for confounders [162]. In another study of 15 chronic hemodialysis patients with bacteremia, Seraph 100 use was associated with a trend towards reduced bacterial load in blood [163]. In both trials, treatment with Seraph appeared well tolerated with infrequent adverse events. RCTs are ongoing in the setting of sepsis and COVID-19-related ARDS (NCT04260789, NCT04547257).

The GARNET^® hemofilter is a second HP device capable of removing pathogens from the bloodstream. This novel device makes use of mannose-binding lectin (MBL), which is a human opsonin that binds multiple carbohydrate patterns present on the surface of bacteria, viruses, fungi, and parasites [164]. The FcMBL is a genetically engineered protein derived from MBL which coats the hollow polysulfone fibers of the GARNET filter. A prospective, multicenter feasibility study in hemodialysis patients with bloodstream infection is underway (NCT 04658017).

Lastly, the Hemopurifier^® filter combines plasmapheresis and adsorption. The adsorptive component is a lectin protein with an affinity for ubiquitous glycoproteins found on enveloped viruses [164]. This adsorbent protein is bound to the extra-capillary space of the filter, where pathogen is immobilized after filtration through 200-nm pores. Data to support the use of Hemopurifier are currently limited to in vitro studies; small safety studies; and case reports in patients with hepatitis C, Ebola, and COVID-19 [165‒168]. A planned prospective trial (NCT04595903) in COVID-19 patients was recently terminated due to inadequate enrollment.

Removal of endotoxin early in gram-negative sepsis has been proposed to reduce the inflammatory cascade that follows. The Toraymyxin^® filter is an adsorptive filter with polymyxin B-immobilized fiber columns [151]. Polymyxin B is an antibiotic capable of binding and inactivating endotoxins but is nephrotoxic and neurotoxic when administered systemically [169]. Preclinical canine studies showed that direct HP with polymyxin B-immobilized fibers resulted in improved survival during endotoxemia [169]. However, multiple human RCTs comparing polymyxin B to standard treatment have yielded mixed results [170‒172]. An additional multicenter RCT (NCT03901807) is ongoing, which, in contrast to prior studies, is specifically targeting subjects with intermediate endotoxin activity (EA) levels (≥0.60 to <0.90 EA units).

CytoSorb^® and oXiris^® are two devices developed for direct cytokine removal. CytoSorb technology utilizes a synthetic polymethylmethacrylate membrane with a symmetric porous structure capable of adsorbing both small- and middle-molecular-weight molecules including cytokines, beta-2-microglobulin, and immunoglobulin light chains [173]. Though retrospective studies of CytoSorb have been promising, prospective trials thus far have yielded no benefit or even signals for harm [174‒178]. One should note that CytoSorb nonselectively removes both anti- and pro-inflammatory cytokines, implying that optimal use requires appropriate patient selection and specific timing of application. Further, CytoSorb is incapable of removing endotoxins [151].

The oXiris filter, in contrast, has been specifically designed to nonselectively remove both cytokines and endotoxins. The oXiris filter is composed of three layers [151]. The outermost layer is composed of a polyacrylonitrile copolymer (AN69) membrane which has a negatively charged hydrogel structure that readily adsorbs cytokines. The middle layer is composed of multiple layers of positively charged polyethyleneimine which adsorbs endotoxin. The innermost layer is a heparin-grafted membrane which serves to reduce thrombogenicity but renders oXiris contraindicated in heparin allergy [151]. In a study comparing the three devices, oXiris was shown to have the cytokine-removal capacity of CytoSorb and the endotoxin-removal capacity of Toraymyxin [179]. Human data to support the use of oXiris have been limited thus far to pilot trials and case series of patients with septic shock, COVID-19, and/or AKI, with most – though not all – studies demonstrating decreases in IL-6 and other cytokines with improvement in intermediary endpoints such as serum lactate levels, organ function, illness severity scores, and vasopressor requirements [180‒184]. RCTs are ongoing in septic shock, cardiogenic shock, and cardiac surgery (NCT04201119, NCT04957316, NCT04997421, NCT05642273). A similar device is the SepXiris^® hemofilter, which has shown to effectively and nonselectively adsorb inflammatory mediators in preclinical studies and has been approved as a treatment for sepsis since 2014 in Japan, though data on use in humans have been limited to observational studies [185‒190].

Finally, rather than aiming for removal of harmful substances, a fundamentally different approach to EBP is to achieve immunomodulation through blood-device interaction. While other extracorporeal immunomodulatory devices have been developed and tested during the pandemic [191, 192], the best studied is the selective cytopheretic device (SCD) [193]. An SCD is a synthetic membrane cartridge that binds and deactivates activated leukocytes, eliminating the source of excess cytokine production. Unlike most hemofilters, the blood path through SCDs is external to fibers, allowing for blood to interact with fibers in a low-shear-stress environment. SCD is employed exclusively with CRRT using RCA, in which the low ionized calcium in the extracorporeal circuit minimizes activation of additional leukocytes. By deactivating activated neutrophils and shifting monocytes towards a reparative phenotype, SCD aims to restore homeostasis. SCD has been evaluated in preclinical and clinical studies of sepsis, AKI, cardiopulmonary bypass, COVID-19, and critical illness, demonstrating signals of benefit on organ function, renal recovery, and survival [193‒201]. The most notable study was a multicenter RCT of 134 patients with AKI requiring CRRT in which, though no difference was seen in the primary 60-day mortality outcome, a reduction was seen in the secondary endpoint of death or dialysis dependency at 60 days among subjects in which the ionized calcium was at the target of ≤0.4 mmol/L for 90% of the treatment time [200]. A recent two-center prospective non-randomized study of 22 patients with COVID-19 demonstrated reductions in IL-6 and other cytokines associated with death in COVID-19, improved oxygenation, and decreased mortality compared to contemporary controls [201]. Additional prospective studies of SCD in cardiorenal syndrome, hepatorenal syndrome, and AKI complicated by other organ dysfunctions are underway (NCT04898010, NCT03836482, NCT04589065, NCT05758077).

Various circuit configurations can be employed to carry out EBP with the above devices. EBP can be performed using a traditional HD machine and an arteriovenous fistula or graft in patients with preexisting vascular access, though central venous access is more commonly obtained with the placement of a dual-lumen catheter. Though separate blood pumps can be utilized, the EBP circuit is usually driven by either an HD or CRRT machine, depending on the hemodynamic status of the patient. Figure 7 displays a non-exhaustive list of configurations that can be potentially utilized for EBP.

Despite all the scientific progress in developing EBP devices, many questions remain regarding the optimal use of these technologies outside of clinical trials. For example, given the heterogeneity of treatment effect of many of these devices and the heterogeneity of sepsis itself, additional biomarkers or other tools to sub-phenotype patients are required to best guide patient selection for EBP [202, 203]. Moreover, studies examining the timing of application of EBP in the ICU are sorely needed. The immunobiology of sepsis is characterized both by periods of immune overactivation, in which overwhelming inflammation drives organ dysfunction – and eventual immune dysfunction – previously described as the compensatory anti-inflammatory response syndrome (CARS) [204]. Likewise, AKI has been associated with a similar state of immunoparalysis characterized by impaired cytokine production and subsequently increased risk of infection [5, 205‒207]. Therefore, application of a device which nonselectively removes both pro- and anti-inflammatory mediators may not be beneficial if applied at the wrong time. For EBP to have a realistic hope of beneficially altering the course of critical illness, additional tools are needed to identify which specific patients at which timepoints in the course of illness will benefit the most. Similarly, prospective studies are needed to elucidate the optimal monitoring approach, the endpoints of therapy, and the timing of discontinuation of EBP [203].

As we work to determine the optimal role of individual EBP devices, we look forward toward a future of EBP which may involve sequential extracorporeal therapy (SET). Based on the pathophysiologic and immunological patterns of sepsis, AKI, and inflammation and the evolving knowledge that organ crosstalk significantly drives multiorgan dysfunction, a rationale for SET has been proposed [208]. For example, SET may involve pathogen removal, followed by endotoxin and cytokine removal and/or immunomodulation, followed subsequently by traditional organ support as needed. However, prospective data to support a SET strategy are currently lacking [151, 208]. Prospective studies to evaluate the differential clinical effects of these devices in isolation and as part of SET are needed.

This review is of two selected topics for the inaugural proceedings of the UAB CRRT Academy: (1) non-invasive hemodynamic monitoring and POCUS to guide fluid management with RRT and (2) evolving technologies of blood purification, highlighting the dynamic nature of the field of critical care nephrology. Furthermore, the provided appraisal of the current literature illustrates the important process of identifying gaps in knowledge to generate needed evidence that could inform the future clinical trials and implementation science necessary to achieve the goal of improving the quality and value of care at bedside in the ICU. The fact that there are more questions than answers in the presented topics underpins the need to nurture critical thinking in trainees, scientists, and practitioners in the field. In this context, the UAB CRRT Academy is committed to iteratively improving its program, diversification, and inclusivity and to the dissemination of the generated content to the broader critical care nephrology community.

J.P.T. reports receiving research funding from La Jolla Pharmaceutical Company; having ownership interest via current or previous stocks or options in Novo Nordisk A/S; and having consultancy agreements with and serving on a speakers bureau for Outset Medical. W.B.S. reports receiving research funding from the Fonds de Recherche du Québec en Santé and the Kidney Foundation of Canada (KRESCENT program) and receiving consulting fees from GSK. K.E. is employed by DaVita Kidney Care. J.Rd.S. is employed by Baxter Healthcare Inc. J.A.N. reports serving on the editorial boards of Advances in Chronic Kidney Disease, American Journal of Kidney Diseases, and Kidney360; serving as a guest editor for critical care nephrology in Advances in Chronic Kidney Disease and a section editor for Clinical Nephrology; and having consultancy agreements with Baxter Healthcare Inc., Biomedical Insights, and Leadiant Biosciences. A.T. reports serving on a speakers bureau for Baxter; having a patent on a 0.5% trisodium citrate solution for CRRT anticoagulation (the license has been bought by Baxter); having consultancy agreements with Baxter Healthcare Inc.; serving on the editorial boards of CJASN and Kidney International; and receiving honoraria from UpToDate. The other authors report having no conflicts of interest to declare.

The authors received no funding relevant to this manuscript.

Conceptualization and supervision: W.B.-S., J.A.N., and A.T.; original draft writing: J.P.T., A.Z., and W.B.-S.; review and editing: J.P.T., A.Z., W.B.-S., J.C., M.J.C., K.E., L.A.J., J.Rd.S., C.W., L.Y., A.B.B., W.M., R.S., K.M.W., J.A.N., and A.T.

J. Pedro Teixeira and Amanda Zeidman are co-first authors; Javier A. Neyra and Ashita Tolwani are co-senior authors.