Automated hemodynamic resuscitation system for computer-controlled norepinephrine, vasopressin, and fluid therapy in endotoxin-induced shock: a feasibility and physiological proof-of-concept study

Background To support hemodynamic management in vasoplegic shock, we previously developed a closed-loop automated infusion system of norepinephrine (NE) and fluid to automatically restore arterial pressure (AP) and cardiac output (CO). To enhance the practical feasibility and efficacy of the system, we integrated a minimally invasive CO monitor and a closed-loop control of vasopressin (AVP) infusion to the system and validated the integrated system in dogs with endotoxin-induced shock. Methods Based on the circulatory equilibrium framework, the system directly controls systemic vascular resistance and stressed blood volume by titrating NE and Ringer’s acetate solution (RiA), respectively, thereby restoring AP and CO. AVP is automatically initiated once infusion rate of NE is increased higher than 1 μg·kg⁻ 1 ·min⁻ 1 . The minimally invasive CO monitor estimates CO (CO es ) by a machine-learning analysis of the peripheral AP contour. In eight anesthetized dogs, the system was implemented, and reference CO (CO ref ) was measured using an aortic flow probe to calibrate CO es and track reference hemodynamics. Endotoxin shock was induced via intravenous lipopolysaccharide infusions. After the target AP and CO es were set to 70 mmHg and 132 (132–133) mL·min −1 ·kg −1 , respectively, the system was activated. Results The lipopolysaccharide infusion significantly reduced AP from 82 (80–85) to 52 (50–56) mmHg and CO ref from 132 (125–146) to 90 (80–103) mL·min −1 ·kg −1 , and increased the blood lactate level (LAC) from 21 (21–23) to 36 (32–38) mg·dL −1 . Once the system was activated, infusions of NE and RiA were automatically initiated. Infusion of AVP initiated during the system control was associated with significant reduction in NE dose while maintaining AP. Within 5 min after system activation, AP and CO es were recovered to >65 mmHg and >122 (122–123) ml·min −1 ·kg −1 , respectively. The median absolute performance errors for AP and CO es were <5%. At 60 min after system activation, AP and CO ref significantly increased to 72 (68–75) mmHg and 130 (127–137) ml·min −1 ·kg −1 and LAC significantly decreased to 26 (25–28) mg·dL −1 relative to the values observed at endotoxin shock. CO es tracked directional changes in CO ref reasonably well. Conclusions This proof-of-concept study validates the feasibility and physiological performance of the system for the hemodynamic management of vasoplegic shock.