Targeted high volume hemofiltration could avoid extracorporeal membrane oxygenation in some patients with severe Hantavirus cardiopulmonary syndrome

Abstract Background Hantavirus cardiopulmonary syndrome (HCPS) has a high lethality. Severe cases may be rescued by venoarterial extracorporeal membrane oxygenation (VA ECMO), alongside substantial complications. High volume hemofiltration (HVHF) is a depurative technique that provides homeostatic balance allowing hemodynamic stabilization in some critically ill patients. Methods We implemented HVHF before VA ECMO consideration in the last five severe HCPS patients requiring mechanical ventilation and vasoactive drugs admitted to our intensive care unit. Patients were considered HVHF‐responders if VA ECMO was avoided and HVHF‐nonresponders if VA ECMO support was needed despite HVHF. A targeted‐HVHF strategy compounded by aggressive hyperoncotic albumin, sodium bicarbonate, and calcium supplementation plus ultrafiltration to avoid fluid overload was implemented on three patients. Results Patients had maximum serum lactate of 8.8 (8.7–12.8) mmol/L and a lowest cardiac index of 1.8 (1.8–1.9) L/min/m2. The first two required VA ECMO. They were connected later to HVHF, displayed progressive tachycardia and declining stroke volume. The opposite was true for HVHF‐responders who received targeted‐HVHF. All patients survived, but one of the VA ECMO patients suffered a vascular complication. Conclusion HVHF may contribute to support severe HCPS patients avoiding the need for VA ECMO in some. Early connection and targeted‐HVHF may increase the chance of success.


| INTRODUCTION
Andes Hantavirus (ANDV) is an orthohantavirus, member of the Hantaviridae family, endemic in Chile and Argentina, and its main reservoir is the long-tailed pygmy rice rat (Oligoryzomys longicaudatus). 1 Humans are infected primarily by the inhalation of aerosolized excreta from infected rodents. 2 Additionally, ANDV is the only Hantavirus known to be transmissible between humans. 3 The incubation period of ANDV varies from 7 to 39 days 4 followed by a cardiopulmonary phase that evolves from dry cough to respiratory failure due to capillary leak into the pulmonary interstitium. Noncardiogenic pulmonary edema is evidenced by chest radiographs showing peribronchial haze and Kerley's B lines that subsequently progress to alveolar flooding with proteinaceous fluid. 5 Hantavirus cardiopulmonary syndrome (HCPS) also includes circulatory shock compounded by hypovolemia and myocardial depression. 6 Using transpulmonary thermodilution (TPTD), we recently documented that increased pulmonary vascular permeability (PVP) was associated to hypovolemia and systolic dysfunction in HCPS patients. 7 Patients with severe HCPS may ultimately develop refractory hypoxemia and/or circulatory shock that leads to death in up to 35-40% of patients, making HCPS is one of the deadliest infectious diseases. [8][9][10][11] Unfortunately there are no drugs with proven efficacy for HCPS and treatment is based on critical care support including judicious fluid management, vasoactive drugs, invasive mechanical ventilation (IMV), and extracorporeal support with venoarterial extracorporeal membrane oxygenation (VA ECMO) in refractory cases. 9,[12][13][14] High volume hemofiltration (HVHF) is a form of depurative therapy that has been used as adjunctive support for refractory septic shock. [15][16][17] As opposed to hemodialysis where depuration is attained by diffusion of small molecules, in hemofiltration, blood is cleared of small and medium-size molecules by convection. 18 The intensity of depuration is set by the filtration or replacement rate; conventional renal replacement rates are 25-35 ml/kg/h, whereas HVHF uses 50-100 ml/kg/h. 19 The exact mechanism of hemodynamic improvement with HVHF in septic shock is unclear and three main mechanisms have been suggested (1) removal of ill-defined vasoactive or myocardial depressant factors 20 ; (2) immune modulation through plasmatic cytokine peak amputation or their mobilization from tissues 19,21 ; and (3) internal homeostasis restoration (acid-base, temperature) devoid of fluid and sodium overload. 22 Despite safety and hemodynamic benefit frequently seen with HVHF in septic shock models 20,23 and clinical case series, [15][16][17]19  For this study, only patients with HCPS admitted to our adult intensive care unit (ICU) and supported with IMV and HVHF were included. The diagnosis of HCPS was confirmed by ANDV specific reversetranscription polymerase chain reaction (RT-qPCR) using an in-house PCR as described in Vial et al. 27 With this technique we quantified viral genome and determined viremia from the buffycoat. In three patients we also performed quantitative enzyme-linked immunosorbent assay (ELISA) detecting ANDV specific immunoglobulin M. Demographic, clinical, laboratory, hemodynamic, and pulmonary monitoring data, as well as hemofiltration and VA ECMO support variables and relevant outcomes, were collected using a standardized case record form.
Anonymized data was then entered into a dedicated database. Data are presented as median and interquartile range (IQR), where appropriate.

| RESULTS
We identified five patients with severe HCPS supported with HVHF before considering VA ECMO support between February and December 2017. These patients are part of the cohort of 11 patients recently reported to describe the TPTD pattern of HCPS. 7 (Tables 1 and 2). They all received hydrocortisone at stress doses with an initial bolus of 100 mg, followed by 50 mg every 6 h. All of them also had an acute respiratory failure with IMV and moderate-to-high positive endexpiratory pressure (PEEP) requirement (Table 2) under deep sedation with continuous infusions of midazolam and fentanyl as well as neuromuscular blockade with cisatracurium. All of them had a severe multiorgan failure, with the highest sequential organ failure assessment (SOFA) score of 12 10-13 points ( Table 1). All patients received an intravenous infusion of convalescent immune plasma at an ANDV neutralizing antibody dose of 5000 U/kg within 15 h of admission according to our local protocol. 10 We did not use normal intravenous immunoglobulin. Individual HVHF settings together with the amount of hyperoncotic albumin, sodium bicarbonate and calcium intravenously added within the first hours of HVHF are shown in Table 3. Individual changes in laboratory and TPTD variables induced after 6 to 15 h of HVHF are shown in Table S1. Patients are presented chronologically.
Despite HVHF, the first two patients continued to deteriorate, developing circulatory failure as attested by declining SI, CI, ITBVI, and rising blood lactate levels while oxygenation worsened and they ultimately required VA ECMO support (Figures 1 and 2, Table S1).
The second patient had a vascular complication during the femoral arterial cannulation for VA ECMO. This resulted in a large superinfected hematoma of the groin that required vascular and reconstructive surgery with a lengthy hospital stay ( Table 1).
The third patient exhibited a dual behavior after HVHF initiation; initially, SI continued to decrease and CI was maintained at the expense of progressive tachycardia and increasing catecholamine doses, and lactate continued to rise ( Figure 1, Table S1). After 9 h of ineffective standard HVHF, a trial of targeted HVHF was started. This modified approach consisted of aggressive fluid resuscitation with hyperoncotic albumin (20%, 400 ml) and sodium bicarbonate (1.4%, 1500 ml and 5.6%, 500 ml) in addition to calcium supplementation while aggressive UF was added to HVHF (Table 3). The target was to rapidly correct hypoalbuminemia and metabolic acidosis while keeping ionic calcemia in the upper normal range and avoiding fluid overload (FO) ( Table 3 and Table S1).
Shortly after this strategy was started, SI increased more than three times, CI rose more than two times and a prominent lactate washout curve followed ( Figure 1, Table S1). Oxygenation impairment, lung edema, and increased pulmonary vascular permeability remained stable through the course of this dual hemodynamic course ( Figure 2). Abbreviations: EVLWI, extravascular lung water index; HVHF, high volume hemofiltration; HR, heart rate; MAP, mean arterial pressure; IQR, interquartile range; ITBVI, intrathoracic blood volume index; LIS, lung injury score (1-4 points); PVPI, pulmonary vascular permeability index; RR, respiratory rate. LIS greater than 2.5 points has been considered diagnostic for the acute respiratory distress syndrome; P/F ratio, arterial oxygen partial pressure to fraction of inspired oxygen ratio. The fourth patient also received supplemental hyperoncotic albumin, sodium bicarbonate, and calcium while aggressive UF was added to HVHF (Table 3). He had a rapid hemodynamic improvement increasing SI with relatively small increases in ITBVI, reducing tachycardia and lactate levels ( Figure 1 and Table S1). This patient had the least pulmonary involvement of this series. He improved oxygenation and slightly decreased EVLWI and PVPI throughout the course of HVHF (Figure 2).
The fifth patient was intubated 24 h before referral to our center and was transferred as a VA ECMO candidate. Upon arrival at our center, he received large intravenous fluid boluses of hyperoncotic albumin (20%, 400 ml), immune plasma (500 ml), and hypertonic sodium bicarbonate (5.6%, 750 ml) together with calcium supplementation while HVHF was started with a net UF rate of 800 ml/h (Table 3). Stroke index, CI, and ITBVI rapidly increased and then stabilized while lactate cleared ( Figure 1, Table S1)   Crowley et al. 12 Responsiveness to HVHF in septic shock has been assessed by catecholamine requirement, cardiac output, and lactate clearance. [15][16][17] We used VA ECMO need as comprehensive criteria of HVHF responsiveness in these patients.
The key question is why some patients improved hemodynamics while others continued to deteriorate, requiring VA ECMO to avoid demise. Nonresponders could be considered more severely ill on the basis of the greatest viremia (patient 1), lower platelet counts (Table 1), higher temperature (Table S1), lower SI (Table 2 and   Table S1, Figure 1), the greatest severity of respiratory failure (patient 1; Table 2 and Figure 2) and the greatest pulmonary edema in the cohort (patient 2; Table 2 and Figure 2). Yet, severity scores, peak PVPI, hemoconcentration, and lactate levels were similar between HVHF-responders and nonresponders. On the other hand, HVHF was given differently to responders and nonresponders; besides an earlier onset (Table 3), responders received targeted HVHF compounded by aggressive resuscitation with hyperoncotic albumin and sodium bicarbonate while calcium was supplemented and UF was added with the aim of normalizing plasma albumin, bicarbonate and ionic calcium while avoiding FO (Table 3 and Table S1). Patient 3 provides a good comparison of the standard and targeted HVHF approaches. Although connected early to standard HVHF, this patient initially seemed to follow the path of hemodynamic deterioration of patients 1 and 2, but when switched to targeted HVHF, SI rose, heart rate dropped and lactate washed out (Figure 1). We saw the same behavior on patients 4 and 5. The whole series seems to suggest a learning curve in terms of both progressively earlier connection and active HVHF optimization to swiftly restore homeostasis.
The first report of hemofiltration in a septic model showed reversal of myocardial depression ascribed to the removal of a filterable cardiodepressant factor. 20 Then HVHF showed improved myocardial performance in an endotoxin-induced shock model. 23 Myocardial depression was reversed only by early hemofiltration in a pneumonia model, suggesting that timing could be important. 28 In a case series of refractory hypodynamic septic shock subjected to HVHF nearly half reversed myocardial depression and survived. 15 Similar to our HCPS series, earlier onset of HVHF was associated with a positive response. 15 Since then, a number of small studies have suggested a favorable effect of HVHF in refractory septic shock, mostly in terms of catecholamine requirement, 16,29,30 but also in terms of oxygenation. 17,31 The dominant explanation for these effects has been the removal of unselected inflammatory mediators. 19,21 Alternatively, some suggest that hemodynamic improvement by HVHF depends on the prompt homeostatic restoration of body temperature and extracellular fluid (ECF) composition. 22 Targeted HVHF could have boosted the benefits of standard HVHF through potential mechanisms given below.

| Correction of metabolic acidosis and calcium supplementation
Metabolic acidosis is known to depress myocardial function 32 and to reduce catecholamine responsiveness in myocardiocytes and vascular myocytes. 33 Acidemic pulmonary vasoconstriction 32 may contribute to acute cor pulmonale in acute respiratory distress syndrome. 34

| Hyperoncotic albumin fluid resuscitation
One answer to the classical critical care conundrum of fluid resuscitation in the context of leaky capillaries is the use of colloids as resuscitation fluids. Artificial colloids have been abandoned due to side effects, and albumin is the only colloid still in use. 36 In septic patients a bolus of 200 ml of 20% albumin expands intravascular volume by 430 ml at 30 min. 37 Besides preload augmentation, albumin has shown positive effects on myocardial contractility in endotoxemic 38 or cirrhotic myocardial depression 39 models. The ALBIOS trial showed that in patients with severe sepsis the use of 20% albumin to maintain serum albumin above 3 g/dl was associated with higher mean arterial pressure and less FO; moreover, the septic shock subgroup showed a survival benefit. 40 In healthy volunteers, 20% albumin produced a larger increase in SI than a fivefold crystalloid bolus while increasing lung diffusion capacity, suggesting balanced salutary hemodynamic and pulmonary effects. 41 This has led to the concept of 20% albumin "small volume resuscitation." 42 The use of large 20% albumin boluses during targeted HVHF may have contributed to the hemodynamic stabilization that averted VA ECMO connection in HVHF-responders. The fifth patient provides a good example; 400 ml of 20% albumin were given, ITBVI increased, SI increased, HR dropped and lactate washed out (Figure 1 and Table S1).

| Fluid overload management
During targeted HVHF not only isovolemic HVHF took place, but large UF amounting to 2.5-5 L was added to give room for bicarbonate, albumin, and immune plasma while sparing FO which has been independently associated with unfavorable outcomes in ICU 43 and HCPS patients. 14 As it was seen on patient 5, fluid resuscitation was followed by an exacerbation of lung edema and respiratory failure that could be contained by high PEEP levels ( Figure 2 and Table S1). Ultrafiltration allows to dynamically balance cardiac preload optimization and containment of pulmonary flooding as achieved in patients 3 and 4 who slightly increased their ITBVI without increasing EVLWI (Figures 1 and 2, Table S1). Targeted HVHF could provide a mechanism to achieve homeostasis while withholding FO until endothelial tight junctions are restored.

| Temperature control
Fever increases oxygen consumption and CO 2 production. 44 Convective therapies produce heat loss 22,45 that could reduce both. 46 In the context of circulatory failure where oxygen delivery is insufficient to sustain tissue demands, temperature control could help to avoid tissue oxygen debt and restore homeostasis. 44 In fact, mild hypothermia improved stroke volume, arterial pressure, mixed venous oxygen saturation, and survival in a cardiogenic shock model. 47 In patients with cardiogenic shock mild hypothermia reduces heart rate and catecholamine requirements while increasing ejection fraction. 48 Additionally, in patients with septic shock cooling decreased vasopressor requirement and short-term mortality. 48 Temperature control could therefore be another potential mechanism of benefit of HVHF. Even though in our series temperature reduction was more prominent in HVHF-nonresponders, absolute temperature reached was lower in HVHF-responders (Table S1).
Severe HCPS remains a condition with high lethality. VA ECMO has decreased mortality from 100% to 33% in the most severe cases. 9 Unfortunately, arterial access, as well as bleeding complica- based on a small sample of an infrequent disease.

| CONCLUSION
Considering the lack of specific therapy for HCPS, the significant morbidity attributable to VA ECMO, and the ease of HVHF in the ICU setting, our report provides relevant data to consider an early targeted