Intravenous albumin for the prevention of hemodynamic instability during sustained low-efficiency dialysis: a randomized controlled feasibility trial (The SAFER-SLED Study)

Hemodynamic instability is a frequent complication of all forms of renal replacement therapy (RRT) commonly used in the intensive care unit (ICU), including sustained low-efficiency dialysis (SLED) [1]. Hemodynamic instability during RRT (HIRRT) is associated with increased in-hospital mortality [2] and may decrease the likelihood of recovery of kidney function to dialysis-independence [3, 4]. In addition, HIRRT can limit ultrafiltration with RRT and perpetuate fluid overload further worsening mortality and reducing the likelihood of kidney function recovery in patients with acute kidney injury (AKI) [2, 5‐7].

Albumin is the primary contributor to intravascular colloid oncotic pressure [8, 9]. Intravenous hyperoncotic albumin may be administered prior to, or during RRT, in order to prevent HIRRT and/or augment ultrafiltration [10]. Theoretically, it works by promoting intravascular plasma refilling from the extravascular compartment to compensate for fluid removal with ultrafiltration. While several small, cross-over trials have reported hemodynamic benefits [11‐13], no trial has been powered for clinically relevant outcomes. Furthermore, intravenous albumin is expensive [14] and not without potential harms [10].

Clinical trials are needed to determine if the administration of intravenous albumin to mitigate HIRRT can improve clinically meaningful outcomes. In order to inform the design and assess the feasibility of such a trial, we undertook a feasibility randomized controlled trial (RCT) comparing intravenous 25% albumin to 0.9% saline during SLED sessions initiated in ICU for patients with AKI.

This single-center feasibility RCT sought primarily to assess the feasibility of a future larger clinical trial that will examine the efficacy of administering intravenous hyperoncotic albumin during SLED sessions to mitigate HIRRT and impact clinically meaningful outcomes. The number of refusals to participate was lower than expected and suggests that there may be fewer objections to participation in a study that involves an intervention that is already frequently applied in the care of critically ill patients and is known to be safe in that context [18]. The use of deferred consent in circumstances in which a surrogate decision-maker could not be reached prior to patients receiving their first SLED session ensured that fewer patients were missed for inclusion than would have been otherwise [19]. Adherence to the protocol was acceptable, meeting the pre-specified target of patients correctly receiving both doses of the assigned intervention (albumin or saline placebo) for at least 90% of SLED sessions overall. The rate of contamination (i.e., open label albumin use was low) but does suggest that a future study should consider refining inclusion criteria or restricting open label albumin use so as to reduce the likelihood of contamination. The pre-specified progression criteria were all achieved and this suggests that a larger trial powered for important clinical endpoints is feasible in this patient population and clinical setting.

While the principal objective of this pilot was to determine feasibility, we observed a number of differences between the groups with respect to blood pressure changes that support the hypothesis that albumin may reduce the incidence of hemodynamic instability among critically ill patients on RRT. Across all completed SLED sessions, albumin administration was associated with 10 mmHg less drop in SBP, on average, compared to saline. This did not appear to have an impact on ultrafiltration volumes, which were not different between the two groups. Notably, there were important baseline differences that could partially account for the hemodynamic findings. In particular, there was more vasopressor use at the start of treatment for patients in the saline arm (80% vs 61%).

Despite evidence of widespread use of hyperoncotic albumin for the prevention or treatment of hemodynamic instability during RRT, few trials have directly addressed its efficacy [10, 20]. Our exploratory hemodynamic findings are consistent with the results of a recent cross-over trial assessing the use of hyperoncotic albumin in hospitalized patients receiving hemodialysis (n = 65; 249 sessions) that included some sessions in ICU and also suggested that hyperoncotic albumin enhanced ultrafiltration [11]. To our knowledge, the only other trial to assess the use of hyperoncotic albumin in critically ill patients with AKI was an 8 person cross-over study that showed improved hemodynamics if 17.5% albumin was used as the priming solution for hemodialysis in septic patients with AKI [13]. Our trial suggests that hyperoncotic albumin may have a significant hemodynamic benefit even in critically ill patients with a high severity of illness on vasopressors and with rapid loss of intravascular albumin via damaged endothelial glycocalyx [10]. Given that it showed benefit in SLED, where ultrafiltration rates are relatively low compared to intermittent hemodialysis, it suggests that, even in patients receiving CRRT, administration of intravenous albumin might potentially improve hemodynamic tolerance of fluid removal.

We observed that significantly fewer patients died in hospital in the albumin group. While we feel that this finding warrants follow-up with a larger RCT, this study’s purpose was to evaluate feasibility and it was not powered to evaluate this outcome. In addition, the observed risk difference is too large to be plausibly explained by the intervention and the relatively small differences in hemodynamic instability that we observed.

There are several strengths of this trial. It was conducted and analyzed in accordance with a previously published protocol and data collection was detailed, including sufficient measures of illness severity, serum albumin, vasopressor use and fluid removal parameters that the observed intra-dialytic blood pressure differences in both study groups could be evaluated within a clear context and according to definitions defined a priori. Another strength is that broad inclusion criteria and the approved use of deferred consent facilitated a higher-than-anticipated recruitment rate. Although adherence was only 90% to the study assignment (lower than expected given that it was carried out by ICU nurses), we determined that the cause of some missed doses was not related to the trial conduct per se but due to the unique circumstance of implementing a new electronic medical record (plus computer order entry) at our institution in June 2019.

There are multiple important limitations of this study. While treatment allocation was blinded, most nurses would likely have been able to determine it since 25% albumin and saline are visually distinguishable when connecting tubing for infusion. Treating physicians likely remained blinded throughout although we did not formally assess this. Another theoretical limitation relates to the use of saline as a placebo in patients who may already be fluid overloaded and require ultrafiltration with RRT. Administration of additional fluid could be harmful as fluid overload is associated with worse outcomes in patients who require RRT [2, 21]. Practically, this is not a major concern: the overall volume of study fluid administered per session is small (200 mL) and is accounted for within the net ultrafiltration goal ordered for each patient’s SLED sessions. Notably, ordered ultrafiltration was achieved for nearly all sessions in both arms of the study suggesting that this intervention (or the placebo) did not contribute to worsening fluid overload. Another limitation relates to the generalizability of our findings. Our study consisted mainly of patients with septic shock. While the patients were enrolled at mixed medical–surgical ICUs at both campuses (with no cardiac surgery patients at either), no post-operative patients were included. Lastly, as a single-center study with a small sample size, as mentioned above with respect to vasopressor use prior to SLED treatments, baseline imbalances in the study groups were present.

In conclusion, this single-center RCT of intravenous hyperoncotic albumin versus saline during SLED treatments in ICU achieved all of its predefined feasibility outcomes, thereby demonstrating that a larger scale trial is feasible. Future trials, adequately powered for outcomes, are needed to determine whether intravenous hyperoncotic albumin has a beneficial impact on HIRRT and relevant clinical outcomes for critically ill patients with AKI requiring RRT.