Perioperative Goal-Directed Hemodynamic Therapy: From Invasive Monitoring To Automated Physiological Closed-Loop Systems

Perioperative goal-directed hemodynamic therapy (GDHT) has evolved from invasive “supra-physiological” maximization of oxygen delivery into minimally and non-invasively guided automated stroke volume optimization. Throughout this evolution, investigators have simultaneously developed novel monitors, updated strategies, and automated technologies to aid them in GDHT implementation. In particular, closed-loop systems have been created to both increase GDHT compliance and decrease physician workload. Currently, these automated systems offer an elegant approach to help the clinician optimize cardiac output and end-organ perfusion during the perioperative period. Most notably, automated fluid optimization guided by dynamic parameters of fluid responsiveness has shown its feasibility, safety, and impact. Making the leap into fully automated GDHT has been accomplished on a small scale, but there are considerable challenges that must be surpassed before integrating all hemodynamic components into an automated system during general anesthesia. In this review we will discuss the potential future of automated GDHT by covering the key events that paved the way from initially complex and time consuming approaches to simple yet effective hands-free strategies.


Background
Modern anesthesia revolutionized medicine and introduced countless technological advances that have greatly improved patient safety. Anesthesiologists are now both specialists in anesthesia delivery and experts in perioperative medicine. Nevertheless, morbidity and mortality still occur in surgical patients. Postoperative complications both negatively impact patient wellbeing and significantly increase healthcare costs [1]. In 2012, an observational study reported that perioperative mortality in Europe remained as high as 4%. The study also underlined the limitations of current perioperative medicine and the need for improved patient care [2]. High-risk patients, defined either by their age, comorbidities, or the surgery itself, make up 15% of the global surgical population and 80% of perioperative deaths [3]. The overwhelming cause of death in these patients is fundamentally due to cellular hypoxia following an inadequate balance between oxygen delivery (Do 2 ) and tissue metabolic demand (Vo 2 ) [4]. Goal-directed hemodynamic therapy (GDHT) is a strategy that aims to correct this imbalance and its associated complications. This strategy first appeared in the 1980s and remains a key topic in perioperative and intensive care medicine today. Over the last decade, automation of fluid delivery for left ventricular preload optimization has shown strong promise while a fully automated GDHT system may soon become a realistic approach for personalizing cardiac output (CO) and end organ perfusion.

From Maximizing Oxygen Delivery To Optimizing Stroke Volume
In 1985, Schultz et al. published a randomized controlled trial (RCT) that demonstrated the benefits of hemodynamic maximization in patients undergoing hip fracture repair [5]. A total of 70 patients were randomized to receive either standard care or preoperative, intraoperative, and postoperative GDHT guided invasively with a Swan-Ganz catheter. Mortality decreased tenfold from 29% in the standard care group to 2.9% in the GDHT group. The same year, another cohort study of 220 critically ill surgical patients determined that patients that survived had improved cardiac function, better pulmonary reserve, lower pulmonary artery pressure, increased Do 2 , and increased Vo 2 despite both survivors and non-survivors initially having vital signs within the normal range. Non-survivors, on the other hand, developed lactic acidosis that was attributed to a defect in oxygen extraction due to microcirculatory alterations [6]. In 1988, Shoemaker et al. demonstrated the detrimental effect of perioperative tissue oxygen debt, defined as the measured Vo 2 minus the estimated Vo 2 requirements. All patients developed tissue oxygen debt during the intraoperative and immediate postoperative periods. While survivors quickly compensated tissue oxygen debt, it persisted and increased in non-survivors [7]. These studies led to the hypothesis that a hemodynamic approach that aimed at maximizing Do 2 would decrease perioperative mortality and morbidity related to tissue hypoxia.
Shoemaker et al. tested this hypothesis in a population of high-risk surgical patients in a RCT that compared standard care with or without invasive hemodynamic monitoring to a Swan-Ganz guided supra-physiological group having the following hemodynamic goals: CI > 4.5 L/min/m², Do 2 > 600 ml/min and Vo 2 > 170 ml/min/m². Patients in the supra-physiological GDHT group had less postoperative complications, shorter intensive care unit (ICU) length of stay (LOS), and reduced mortality. In 1993, Boyd et al. confirmed this hypothesis in a RCT of 107 mostly surgical trauma patients [8]. They showed once again that maximizing patient Do 2 with a supra-physiological GDHT strategy in the preoperative, intraoperative, and postoperative periods led to decreased morbidity and mortality. In a follow-up study of these patients, the authors noted that long term survival was also greater in the supra-physiological GDHT group [9]. Shoemaker and his team would then attempt to determine, on the one hand, which patients would benefit the most from this strategy [ . A better approach has since been described that aims to limit fluid infusion so as to optimize, and not maximize, SV by using an approach that adapts to the patient's fluid responsiveness [27]. In the following years, anesthetist would simplify these strategies to focus mainly on CO optimization, while a concomitant paradigm shift would push fluid administration away from the arbitrary concept of "restrictive versus liberal fluid therapy" towards a "goal-directed" strategy. volume-to-volume compensation with a colloid), when compared to what at that time was standard care (i.e. 3-7 ml/kg/h crystalloid third space loss compensation with 1000-1500 ml of crystalloid for up to a 500 ml loss followed by colloid infusion for greater losses), was associated with decreased postoperative morbidity [29]. More recently, however, Myles et al. showed in a large trial of over 3000 patients that there was no difference in long term outcome when comparing "restrictive versus liberal" approaches and that a restrictive approach could even be associated with a higher rate of acute kidney injury in high-risk patients during major abdominal surgery [31].
This fluid management controversy, which has spanned for over two decades, is in large part due to the lack of clear definitions of "restrictive" and "liberal". For example, the "restrictive" and "liberal" fluid regimens were completely different in the above studies. These subjective definitions thus depend on the arbitrary threshold each study design sets! A better approach would be to look at the question from another perspective: would it not be better to optimize the patient's cardiac preload by using an individualized goal-directed fluid strategy?

Heart-lung Interactions: Providing Reliable Goals For Fluid Therapy And Preload Optimization
Goal-directed therapy was initially limited by the invasiveness of pulmonary artery catheter monitoring [5,7,10]. Guiding fluid therapy based on transesophageal Doppler provided a much less invasive approach, but adoption was limited by its considerable learning curve. With the introduction of pulse contour technology derived from the heart-lung interactions, semi-invasive and non-invasive hemodynamic monitors became available for a broader application of GDHT both in the ICU and operating room [32,33]. During mechanical ventilation, cyclic increase in intrathoracic pressure changes left ventricular preload and afterload. This pressure opposes cardiac venous return and reinforces systemic arterial blood flow away from the heart and out of the thorax. During expiration, the decrease in intrathoracic pressure has the opposite effect. This reversed pulsus paradoxus due to positive pressure ventilation is more prominent in hypovolemic patients and progressively decreases with correction of hypovolemia [34]. Once the patient's preload reaches the plateau of the Frank-Starling curve variations in blood pressure and flow due to heart-lung interactions become minimal.
The dynamic parameters derived from heart-lung interactions include pulse pressure variation (PPV), stroke volume variation (SVV), and Pleth Variability Index (PVI). All have been proposed as useful tools to guide preload optimizing strategies [34][35][36]. These dynamic parameters have been evaluated in multiple studies and are today an essential part of GDHT [37][38][39]. Even a new smartphone application, called Capstesia™ (Galenic App, Vitoria-Gasteiz, Spain), which automatically calculates PPV from a digital picture of the invasive arterial pressure waveform from any monitor screen, has been developed and tested to assess its ability to predict fluid responsiveness or decision making regarding fluid therapy [40][41][42]. This promising smartphone technology needs to be further evaluated to determine its potential role. Although heart-lung interactions provide a means for evaluating fluid responsiveness, several conditions, in addition to mechanical ventilation, are needed for the monitor to provide valid information. For pulse contour analysis, patients must have sinus rhythm, at least 8ml/kg (ideal body weight) of tidal volume for validated thresholds to be accurate, and a heart rate to respiratory rate ratio greater than 3.6. Arrhythmias, aortic regurgitation, sternotomy, thoracotomy, Knowledge and compliance are consequently the main limitations for applying GDHT. One approach that may facilitate adoption would be to automate therapy using a closed-loop or open-loop system.
An automated system capable of administering GDHT can be seen as an extra pair of eyes and hands.
The system observes the patient's hemodynamics, via a monitor, and then acts to maintain predefined goals, often via an automated infusion pump. This in no way exempts the clinician from the requirement of understanding hemodynamic variables and GDHT. The clinician will still need to determine appropriate targets, for example, but it does assure consistent compliance and limits the learning curve. The clinician is then free to focus on more important perioperative tasks that require human intelligence such as reasoning, problem solving, and making critical decisions while the automated system consistently adjusts hemodynamics. Figure 1 depicts the evolution of perioperative GDHT over the past 40 years.

Increasing Compliance With Automation: The Example Of Fluid Therapy
Protocol adherence is essential to any goal-directed strategy, but consistent application is fastidious and does not seem to be feasible in the contemporary intraoperative setting where a single anesthetist is responsible for many tasks. An automated system offers an elegant way to integrate hemodynamic parameters, increase protocol compliance, and free the clinician to focus on other key intraoperative tasks. Closed-loops consist of a controller that monitors at least one parameter and automatically intervenes to maintain a predefined goal. They have been applied in medicine to  [78]. In addition, when compared to manual goal-directed fluid therapy, this system maintains patients in a preload independent state for a longer period of time [81,82]. The advantages of this system should not be ignored. It consistently gives fluids according to well established endpoints for any patient regardless of when or where it is used. This is not the same for anesthetists, who have been shown to be much less consistent in applying hemodynamic protocols.

Maintaining Perfusion Pressure With Closed-loop Vasopressor Infusion: The Next
Step In Automation Blood pressure control has challenged anesthetists for decades. Intraoperative hypotension is associated with poor end organ perfusion and results in increased patient morbidity and mortality [83]. To this day there is still no consensus on the exact definition of intraoperative hypotension and the best perioperative blood pressure goals to target. Futier et al. demonstrated that, in moderate-tomajor abdominal surgeries, an individualized approach that maintained blood pressure within 10% of the patient's baseline value reduced the risk of postoperative organ dysfunction [84]. Maintaining an average predetermined value, however, is not the only challenge in blood pressure optimization. As high systolic blood pressure variability has been associated with increased postoperative mortality and renal failure [85], an effective vasopressor strategy should not only maintain a predetermined blood pressure value but also assure low variability. This requires frequent repetitive tasks and an automated system could outperform clinicians in maintaining a blood pressure target with minimal variability [86].
Several teams have developed closed-loop vasopressor systems. Ngan Kee et al. tested a closed-loop vasopressor system that administers phenylephrine based on non-invasive systolic blood pressure measurements [87]. The developed system has been extensively tested during obstetric anesthesia and, when compared to manually titrated phenylephrine, has been shown to provide better blood pressure control for hypotension associated with spinal anesthesia [88]. Sironis (Irvine, USA) has also recently developed another automated closed-loop vasopressor system that has extensively been tested in simulation [89,90] The future of closed-loop GDHT will probably consist of interdependent systems capable of interpreting, integrating, and optimizing the main determinants of CO and organ perfusion. Such systems do not exist yet, but some recent work has demonstrated that the simultaneous use of multiple physiological closed-loop systems operating in parallel is feasible (Figure 4) [97, 98].
Although automation increases GDHT compliance and may improve outcome in high-risk patients, several potential pitfall must be acknowledged. An automated system should never be left to itself or under the supervision of someone unable to understand the fundamental aspects of anesthesia, intensive care, hemodynamics, perfusion pressure, monitoring, pharmacology, and GDHT. Closed-loop systems should have safety cut-off points if a loss of input (e.g. monitoring dysfunction) or an inability to apply treatment (e.g. pump failure) occurs. The controller itself must have physiologic norms of CO and blood pressure that, if interpreted as inadequate or excessive, lead to treatment modifications.
Despite these safety controls, an error in input, such as the arterial line pressure cell falling to the ground, would lead to an erroneous signal being interpreted as correct. In this situation of an increased hydrostatic pressure difference between the heart and the arterial line transducer, the controller would consider this erroneously high invasive blood pressure as adequate, decrease the vasopressor infusion, and the patient would suffer the negative effects of hypotension. An underestimation of blood pressure, due to an inappropriately elevated position of the arterial line transducer, could lead to excessive vasopressor infusion, hemorrhage, heart failure, and even death.
Another source of potential danger that requires oversight is the closed-loop interventional component. Although drugs with similar effects could probably be interchanged and moderate dose differences in vasopressor or fluid composition would be corrected by the controller, administering the wrong drug, by switching the dobutamine and noradrenaline syringes for example, would lead to devastating consequences. Even if automated, GDHT will always require intelligent and knowledgeable oversight. Supplementary safety procedures, such as barcode recognition by the closed-loop pump of pre-filled syringes could be useful, but physicians remain the fundamental safeguard of these systems.
Perioperative GDHT has evolved from invasive "supra-physiological" maximization of oxygen delivery to minimally and non-invasively guided automated stroke volume optimization. Closed-loop systems provide an elegant approach for increasing compliance and implementing adaptive therapies that individualize treatment. Integration of multiple closed-loop setups into a universal interdependent system could represent the next step in optimizing end organ oxygen delivery. Finally, even with these advances in automation, physician oversight will remain essential to guarantee proper hemodynamic goals and assure perioperative safety.