Continuous Venovenous Hemofiltration is Associated with Improved Survival in Burn Patients with Shock: A Subset Analysis of a Multicenter Observational Study

Acute kidney injury (AKI) significantly complicates the hospital course of critically ill patients. Increased utilization of renal replacement therapies (RRTs) and advancements in the technology has reduced mortality [1, 2]. Despite improved outcomes, there is a lack of consensus regarding many aspects of RRT to include modality, timing of initiation, and dose [3-9]. Varying trial designs and target populations make direct comparison of studies difficult [10, 11]. Ultimately, the choice of modality often comes down to prescriber preference and availability.

In patients with severe burn injuries, development of AKI is associated with longer length of stay and higher mortality [12-14]. Early choice of continuous venovenous hemofiltration (CVVH) over traditional dialysis was identified to improve survival and reduce vasopressor requirements in severe burn patients with AKI [2]. During the RESCUE trial, use of high-volume hemofiltration (HVHF), 70 mL/kg/h, in burn patients with shock reduced 48-h vasopressor requirements and 14-day Multiple Organ Dysfunction Score, as compared to standard doses (20–35 mL/kg/h) [5]. Concurrently, a multicenter observational study was performed to evaluate RRT practices and outcomes among severely injured burn patients [3]. We found centers initiated therapy early, preferred CVVH, and demonstrated reduced mortality and similarly better full renal recovery rates, as compared to other studies [15-21]. The purpose of this subset analysis of the study cohort was to determine if modality characteristics of RRT impacted survival.

Multilevel, Institutional Review Board approval was obtained to conduct this multicenter, observational study. Ten burn centers participated from February 2012 to February 2016. All patients or their legal authorized representative gave consent prior to enrolling in the study.

The observational study enrolled patients placed on RRT for any indication, but not enrolled in the concurrent RESCUE trial (NCT01213914). All adult patients 18 years or older, admitted with severe burns, and placed on continuous RRT for acute indications were included. Patients were excluded if they were on dialysis for ESRD on admission, prisoners, already enrolled in the RESCUE trial, or had non-thermal injuries.

After informed consent was obtained, data were collected prospectively. For patients already started on RRT prior to approval of this concurrent enrolling study, data were collected retrospectively. Patient demographics, injury characteristics, laboratory data at time of RRT initiation, RRT treatment data, and outcome data were collected. Data were entered into a web-based electronic data management portal (Velos, Inc., Fremont, CA, USA). Data related to RRT treatment included mode of therapy, ordered and delivered therapy dose, AKI stage at initiation (KDIGO criteria), prescribing provider specialty, and method of regional anticoagulation. Outcome data included in-hospital mortality and need for RRT at discharge. A 6-month follow-up audit was performed to determine which survivors remained on RRT.

Analysis was performed on de-identified data. χ² test was utilized for categorical comparisons. Continuous variables were compared using one-way ANOVA. Demographics were expressed as means with standard deviations or medians with minimum and maximum ranges, depending on normality determined via Shapiro-Wilk test. Sample size determinations were calculated utilizing χ² test with a Yates Correction Factor to attain at least an 80% power with an alpha of 0.05. Kaplan-Meier plots were utilized to assess survival impact over time. SPSS 22.0 was utilized for statistical analysis.

During the study, data from 171 enrolled patients were collected for analysis. One patient was excluded from the analysis for incomplete data. Patients in the cohort had a mean age of 51 ± 17.4 years, and 80% were male. The mean total body surface area burned was 38% with 24% being full thickness. Thirty-four percent sustained inhalation injuries. The mean Acute Physiology and Chronic Health Evaluation II score was 26. Ninety-one patients (54%) were treated with CVVH, 53 (31%) with continuous venovenous hemodiafiltration, and 26 (15%) with some form of dialysis. Twenty-three patients were started on continuous venovenous hemodialysis, 2 on intermittent hemodialysis, and 1 on sustained low-efficiency dialysis. Demographics and treatment characteristics were not significantly different between groupings, based on modality (Table 1). Comparisons between survivors and non-survivors can be seen in Table 2. Non-survivors were significantly older with large burns and more likely to sustain an inhalation injury. Survivors were significantly less likely to require vasopressors. Likely related to longer lengths of stay, survivors were also more likely to acquire a ventilator-associated pneumonia.

When comparing patients treated with CVVH versus all other modalities, there was no statistically significant difference in survival; however, the analysis was underpowered (56 vs. 43%; p = 0.124; power = 0.319) (Fig. 1). To adequately detect the presence of the 13% difference in mortality with an 80% power, 546 patients would be required. Survival was improved in the subset of patients receiving vasopressors (n = 77) at the time of RRT initiation (54 vs. 37%; p = 0.032) (Fig. 2). For patients sustaining inhalation injury, there was no difference in survival (53 vs. 47%; p = 0.638; power = 0.071). There was no statistical survival difference in patients developing acute lung injury or acute respiratory distress syndrome (ARDS) (Fig. 3).

There was no difference in survival when comparing patients treated with some form of convection to diffusion alone (53 vs. 31%; p = 0.055; power = 0.47). Three hundred fifty-one patients would be needed to be meet 80% power. Parallel to sample attrition, power significantly decreased when analyzing patients requiring vasopressors (n = 77). Survival difference did not meet statistical significance (47 vs. 18%; p = 0.061; power = 0.571) but would require 167 patients to be adequately powered. There was a difference in survival for patients sustaining inhalation injury, as there were unfortunately no survivors in the diffusion group (42 vs. 0%; p = 0.005) (Fig. 4).

Therapy dose was analyzed based on delivered dose, comparing ≥35 mL/kg/h (HVHF) to standard doses. There was no statistical difference in overall mortality (55 vs. 35%; p = 0.131; power 0.308) (Fig. 5). To give proper chance to detect the observed 20% difference in mortality, 257 patients would be needed. For patients receiving vasopressors, HVHF did not provide a statistically significant survival difference (54 vs. 37%; p = 0.241; power = 0.201). Assuming only a 15% survival benefit, 485 patients would need to be enrolled to be adequately powered. Unfortunately, only 23 patients with inhalation injury had complete data regarding delivered dose, making it difficult to compare survival. Survival failed to meet statistical significance for patients with inhalation injury receiving some form of HV convection (n = 4) versus all other modalities and lower doses (n = 19) (75 vs. 21%; p = 0.067; power = 0.561).

This is the first multicenter cohort study to analyze outcomes based on RRT modality in subsets of burn patients. CVVH demonstrated improved survival in patients requiring vasopressors. Although this observational study does not elucidate the mechanism, the use of convection appears to have benefit in subsets of the burned population, especially in patients requiring vasoactive medications.

Conflicting data exist as to the impact of modality choice on outcomes in critically injured patients with AKI [10, 18, 21-26]. However, a high heterogeneity exists between studies, making generalization and conclusions difficult. This study is the first to analyze mortality in burn patients requiring RRT. While this study was underpowered for some of the subgroup comparisons, the divergence of outcomes is noteworthy. The data within should be helpful for future investigators.

Utilizing convection benefited patients with burn injuries in this study. While cytokine removal has been the theorized mechanism leading to enhanced hemodynamic stability, clinical data supporting this hypothesis remains conflicting [5, 27-31]. In the RESCUE study, patients treated with HVHF had a faster resolution of their shock, but it did not appear related to removal of interleukin (IL) 6, IL8, IL10, and IL12, interferon-γ, or tumor necrosis factor-α. Of the cytokines chosen to study, none varied over the 48-h measurement period. It is possible different cytokines could be influential or alternative mechanisms could be responsible for the improvement (e.g., metabolic control). Recently, You et al. [32] showed significant removal of cytokines through early application of HVHF in burn patients with sepsis, especially in a subset of patients with very severe burn injuries (>80% total body surface area). The authors also found improved ratios of arterial oxygen partial pressure to fraction of inspired oxygen, improved immunologic function, decreased incidence of sepsis, and improved survival. HVHF was initiated within 3 days postburn and circuits changed routinely at 24 h regardless of clotting. It is unclear if differences existed in daily fluid balances and, if so, could have impacted the findings. It is likely the ideal marker of most benefit has yet to be identified.

Overall mortality was 50% in the observational study and 62% in the interventional study. While still poor, the improvement is noteworthy and consistent with the outcomes trended through the years [12]. In their meta-analysis, Brusselaers et al. [12] noted a median mortality of 80% for patients with AKI requiring RRT. However, there was a significant improvement in outcomes trended from 1950 to 2010.

Although underpowered, HVHF demonstrated a 20% decrease in overall mortality in this study, which conflicts with recent literature on the use of higher doses in critically ill patients. A recent systematic review (4 studies) concluded with a low-level recommendation against the use of HVHF over standard doses but stated the inconsistencies in data and significant heterogeneity across studies resulted in low quality of evidence with more studies needed [33]. Conversely, a large meta-analysis of 21 studies reported a significant survival benefit with use of HVHF, especially in the subsets of patients with sepsis and ARDS [34]. The authors assessed for 2 levels of bias and judged none of the included studies to be at low risk of bias.

Perhaps, the “tip of the iceberg theory” should be revisited [28, 35]. Measuring mediator concentrations in the plasma may not necessarily adequately represent concentrations throughout other multiple compartments. Additionally, simply measuring concentrations of cytokines may not be enough, as substantial interpatient variability exists regarding receptor expression and mediator affinity. It may be enlightening to measure and compare mediator differences between multiple compartments within the burn patient. New theories are shifting focus toward beneficial effects attributed to damage-associated molecular pattern (DAMP) removal, but may require the use of an alternative membrane [28, 36, 37]. DAMPs are released after burn injury and have been linked with inflammation and monocyte activation [38]. Simmons et al. [39] demonstrated the presence of DAMPs in reportedly apheresed, leukocyte-reduced transfusion products, and their link to clinical outcomes [40]. Blood product administration is frequently a component of shock resuscitation in burn patients, but was not collected in this cohort [41].

This study was limited by its observational design. Also, data collection occurred both prospectively and retrospectively, based on approval timing of the observational, parallel randomized controlled trial. The likely presence of bias toward prescribing CVVH in the experienced participating burn centers should be noted. Center effect cannot be excluded, as centers can be biased toward treatment preference. AKI stage was collected, but specific indication was not and is a limitation. Delay in initiation was also much higher than some studies in burn patients showing benefit [8, 32]. The analysis is limited to the data chosen to be collected, and potential exists for other data to be of value. While history of ESRD requiring dialysis was excluded, presence of other stages of chronic kidney disease was not collected. Specifics regarding management of each case of shock were not collected and could have influenced the results. Similarly, this study was not designed to answer if removal of endotoxin could play a role in survival. While mean prescribed therapy dose was 42 mL/kg/h and evenly divided between HV and standard doses, delivered dose was not recorded for 73 patients. Due to incomplete data for some subsets, much of the analyses suffer from being underpowered. Baseline characteristics were not different between modality subsets (e.g., revised Baux score); however, there is a chance a difference still exists and a type II error present. The authors hope the power and sample size analysis will aid future researchers in developing studies to answer much needed questions in this population. However, for this multicenter study, the investigators predicted enrollment of 540 patients over the 4-year study period, demonstrating the difficulty and likely unfeasibility of much larger trials in this population.

This observational study is the largest cohort to date to perform a subset survival analysis of burn patients started on RRT. This study adds supportive survival data for use of convection-based clearance in septic burn patients, although definitive conclusions cannot be drawn from this study alone. The exact mechanism responsible for the possible benefit remains to be elucidated. Convection may benefit patients with extreme inflammatory pathogenesis, such as burn patients with inhalation injury and acute lung injury/ARDS. More research is needed in high-risk acutely ill populations to delineate why CVVH provides a survival benefit for burn patients with AKI and shock.

The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of the Army or the Department of Defense.

The authors would like to specifically acknowledge tremendous work done by all the RESCUE investigators and support staff: Elsa Coates, Erica A. Pamplin, Cathy A. Rauschendorfer, Sonya Charo-Griego, Victoria D. Hatem, Javance R. Tercero, and Linda S. Welker, United States Army Institute of Surgical Research; Thanh Tran and Kristina R. Bolling, MPH, University of South Florida Tampa General Hospital; Marcia Halerz, Loyola University Medical Center; Amber McIntire and Kiera Nelson, Doctors Hospital Joseph M. Still Burn Center; Karen J. Richey and Claudia Islas, Arizona Burn Center Maricopa Integrated Health Systems; Christina Howard, Peter Gales, and Agnes Burris, University of Texas Southwestern Medical Center. The following is a list of all the ABA RESCUE Investigators listed by clinical site: Booker T. King, MD, Julie A. Rizzo, MD, Jeremy C. Pamplin, MD, Ian R. Driscoll, MD, Evan M. Renz, MD, Jonathan B. Lundy, MD, Elizabeth Mann-Salinas, PhD, and Leopoldo C. Cancio, MD, United States Army Institute of Surgical Research; Carl W. Cruse, MD, Christopher A. McFarren, MD, Kimberly S. Brown, ARNP, David J. Smith Jr., MD, and Rachel A. Karlnoski, PhD, University of South Florida Tampa General Hospital; Arif Showkat, MD, Lekha George, MD, Aneel Kumar, MD, Barbara Birmingham, CRNP, Angela L. Arnold-Ross, RN, and Mary E. Bruce, PA, University of Tennessee Firefighters’ Regional Burn Center; Michael J. Mosier, MD, Arthur P. Sanford, MD, David J. Leehey, MD, and Peggie F. Conrad, Loyola University Medical Center; Bruce Friedman, MD, Amy M. Sprague, MD, Robert F. Mullins, MD, Zaheed Hassan, MD, and Joseph R. Shaver, MD, Doctors Hospital Joseph M. Still Burn Center; Kevin N. Fosters, MD, Michael D. Peck, MD, Daniel M. Caruso, MD, and Marlene Albrecht, RN, Arizona Burn Center Maricopa Integrated Health Systems; Steven E. Wolf, MD, Brett D. Arnoldo, MD, Herb A. Phelan, MD, and Ramesh Saxena, MD, University of Texas Southwestern Medical Center; Laura S. Johnson, MD, and Melissa M. McLawhorn, RN, Medstar Washington Hospital Center. Additionally, we are appreciative of the continuous support from the ABA central office: Kimberly A. Hoarle, Janet Turner, Kitty Vineyard, Susan Browning, and Lori Palfalvi; ABA Multi-Center Trials Group: Jeffrey Saffle, MD, Salt Lake, UT, and James H. Holmes, MD, Winston-Salem, NC: the Data Coordinating Center, University of California Davis Medical Center, Sacramento, CA: MaryBeth Lawless (Director), Silvia Hughes, and Katrina Falwell.

IRB approval was granted, and informed consent was given by each patient.

The authors declare to have no conflicts of interests.

This work was funded by the United States Army Medical Research and Material Command (Grant No. W81XWH-09-2-0194).

All authors made substantial contributions to the conception, design, and acquisition of data. D.M.H., J.K.A., and K.K.C. made substantial contributions to analysis and interpretation of data. D.M.H. drafted the manuscript. All authors were involved in revising the manuscript. All authors gave approval of the final manuscript to be published.