Fluid expansion improve ventriculo-arterial coupling in preload patients: a prospective observational study

Methods Arterial output (CO), E A , and E LV were measured before after with of Fluid responders were dened as with V-A coupling the A This prospective, observational study was performed in the Amiens University Hospital cardiothoracic ICU (Amiens, France) over one year. The main inclusion criteria were as follows: age 18 or over, controlled positive ventilation, and a clinical decision to perform FC for volume expansion within the rst hours of to ICU. Exclusion criteria were permanent arrhythmia, cardiac conduction block, pacemaker (or need for temporary pacemaker using epicardial wires), norepinephrine, epinephrine or dobutamine, poor echogenicity, aortic regurgitation, and right heart dysfunction. The indications for FC were arterial hypotension (systolic arterial pressure (SAP) less than 90 mmHg, or mean arterial pressure (MAP) less than 65 mmHg), or SV change greater than 10% during a passive leg raising manoeuvre, or clinical signs of hypoperfusion (skin mottling, and capillary rell time greater than 3 sec).


Results
Twenty-three (77%) of the 30 patients included in the study were uid responders. Before FC, responders had higher mean E A and E A /E es ratio. FC signi cantly increased mean arterial pressure, LVEF, SV and CO, and signi cantly decreased SVRi, E A and consequently the E A /E es ratio. A pre-challenge E A /E es ratio greater than 1.4 was predictive of uid responsiveness (area under the curve [95% con dence interval]: 0.84 [0.66-1]; p < 0.0001).

Conclusions
In FC responders, V-A coupling was characterized by a higher pre-challenge E A /E es ratio (due to a higher E A ). FC improve V-A coupling ratio by decreasing E A but not E es .

Background
Fluid challenge (FC) is the most commonly performed bedside haemodynamic intervention in critical care medicine. In conventional haemodynamic analysis, cardiac output (CO) is considered to be a continuous function, and the heart and vascular system are considered separately. Two different concepts have therefore been developed. In the 1950s, Guyton et al. considered the heart to be a pump driven by continuous ow from a purely resistive circuit -despite the pulsatile nature of this ow (mean atrial pressure -right atrial pressure = CO x systemic vascular resistance (SVR)) (1). Several authors subsequently developed a model of stroke volume (SV) based on the pressure-volume relationships of the ventricle and the vascular system (2)(3)(4). This model considers left ventricular (LV) energetics, myocardial function and ventricular performance by taking into account the interaction between the ventricle and the vascular system. Hence, LV end-systolic elastance (E es ) corresponds to LV contractility and arterial elastance (E A ) corresponds to the effective elastance of the arterial system (2)(3)(4). Given that the ventricular and arterial systems operate simultaneously, ventricular-arterial (V-A) coupling (i.e the E A /E es ratio) determines the SV and ejection pressure (i.e. arterial blood pressure) (2,4). The V-A coupling model has been used in cardiology and cardiac surgery to describe and characterize pathophysiological mechanisms and to evaluate treatment effects (5,6). Recently, we demonstrated that V-A coupling was improved by norepinephrine infusion and that V-A coupling ratio was associated with SV increase (7).
Most of the studies on FC published in the literature have evaluated its effects according to Guyton's model (8,9). This model has provided researchers with a comprehensive overview of the effects of FC (8,9). However, few studies have focused on FC from the perspective of the V-A coupling model (10).
Because treatment of acute circulatory failure comprises several medications ( uid infusion, inotropic or vasopressor use), it would be of interest to know the effect of each treatment on V-A coupling. The clinical relevance of this model is based on the fact that E A /E es predicts outcomes independently from other parameters in patients with cardiovascular diseases (5). A description of the cardiovascular effects of uid expansion may improve our understanding of the pathophysiology of haemodynamic states.
The main objective of this study was therefore to evaluate the impact of FC on V-A coupling, and its determinants. The secondary objective was to determine the value of the pre-challenge E A /E es ratio as a predictor of a post-challenge increase in SV.

Ethics
The study's objectives and procedures were approved by the local independent ethics committee (Comité de Protection des Personnes Nord-Ouest II, Amiens, France; RNI2014-39) on November 26th, 2014). All patients received written information about the study and provided their verbal consent to participate. The present manuscript was drafted in compliance with the STROBE checklist for cohort studies (11).

Patients
This prospective, observational study was performed in the Amiens University Hospital cardiothoracic ICU (Amiens, France) over one year. The main inclusion criteria were as follows: age 18 or over, controlled positive ventilation, and a clinical decision to perform FC for volume expansion within the rst hours of admission to ICU. Exclusion criteria were permanent arrhythmia, cardiac conduction block, pacemaker (or need for temporary pacemaker using epicardial wires), norepinephrine, epinephrine or dobutamine, poor echogenicity, aortic regurgitation, and right heart dysfunction. The indications for FC were arterial hypotension (systolic arterial pressure (SAP) less than 90 mmHg, or mean arterial pressure (MAP) less than 65 mmHg), or SV change greater than 10% during a passive leg raising manoeuvre, or clinical signs of hypoperfusion (skin mottling, and capillary re ll time greater than 3 sec).
Transthoracic echocardiography (with the CX50 ultrasound system and an S5-1 Sector Array Transducer, Philips Medical System, Suresnes, France) was performed by a physician blinded to the study outcomes. Left ventricular ejection fraction (LVEF), end-systolic volume (ESV), and end-diastolic volume (EDV) were measured using Simpson's method on a four-chamber view. The aortic velocity-time integral (VTIAo), preejection time and systolic time were measured by pulsed Doppler at the left ventricular out ow tract on a ve-chamber view. Stroke volume (SV; mL) was calculated as VTIAo × SAo, and was expressed as indexed SV (SVi) = SV/body surface area (ml.m − 2 ). Cardiac output (CO) was calculated as SV × heart rate (HR), and was expressed as indexed CO (CI) = CO/ body surface area (ml min − 1 m − 2 ). Mean echocardiographic parameters were calculated from ve measurements (regardless of the respiratory cycle) and analysed retrospectively.
Left ventricular end-systolic elastance, arterial elastance, ventricular-arterial coupling E es , an index of ventricular contractility, was evaluated by using the noninvasive, single-beat method described by Chen et al. (12). This method is based on the assumption that time-variation of LV elastance is not in uenced by loading conditions or heart rate. E es was calculated by the formula: E es = (Pd -(E Nd(test) * Pes * 0.9)) / (SV * E Nd(test) ). E Nd(test) was obtained from a group-averaged normalized elastance curve value at this same time td (E Nd(avg) ), baseline LVEF and the ratio of diastolic to systolic arterial pressure (Pd / Pes ) (14). E Nd(avg) was determined by a seven-term polynomial function that includes the ratio of pre-ejection period to total systolic period (12). We calculated the coe cient of variation (CV), precision and least signi cant change (LSC) for E es in the rst ten patients. CV was 7.7% ± 0. 6  where ESP is 0.9 x systolic arterial pressure (SAP) (13), E A(MAP) as MAP/ SV (mmHg ml − 1 ) (14), and E A(R/T) as total peripheral resistance / cardiac cycle (mmHg ml − 1 ) (3). Arterial pressure was measured by an invasive radial artery approach. In healthy men and women, mean E A /E es , E A , and E es values measured invasively at rest are 1.0 ± 0.36, 2.2 ± 0.8 mmHg.ml − 1 , and 2.3 ± 1.0 mmHg.ml − 1 , respectively (15)(16)(17). An abnormal E A /E es ratio was de ned as a value greater than 1.36 (17).
The total energy generated by each cardiac contraction is called the "pressure-volume area" (PVA), corresponding to the sum of the external mechanical work exerted during systole (SW) and the potential energy (PE) stored at the end of systole: PVA = SW + PE (18). SW is calculated as ESP x SV. PE is calculated as ESP x ((ESV-V 0 )/2), and assumes that V 0 is negligible compared to ESV. The SW/PVA ratio corresponds to the mechanical e ciency of converting the total mechanical energy (PVA) available to the LV SW (18).

Study procedures
The following clinical parameters were recorded: demographic, ventilation parameters, and primary diagnosis. After an equilibration period, capillary re ll time (measured at the distal phalanx of the index nger), HR, SAP, MAP, diastolic arterial pressure (DAP), central venous pressure (CVP), SVi, CI, EDV, ESV, pre-ejection time, systolic time interval, and blood gas levels were measured at baseline. In the present study, FC always consisted of a 10-minute infusion of 500 ml of lactated Ringer's solution (20). A second set of measurements was performed immediately after FC. All patients were mechanically ventilated in volume-controlled mode with a tidal volume set at 7-9 ml kg − 1 ideal body weight, and a positive endexpiratory pressure (PEEP) of 5-8 cmH 2 O. Ventilator settings were not modi ed during the study period.

Statistical analysis
The sample size was calculated on the reproducibility initially measured in the study reported by Chen et al (12). With a reproducibility of 20%, we calculated that a sample of thirty patients would be su cient to demonstrate an absolute change of more than 20% in the E A /E es ratio in response to FC. The distribution of the variables was assessed by a Shapiro-Wilk test. Data are expressed as number, proportion (in per cent), mean ± standard deviation (SD) or median [interquartile range (IQR)], as appropriate. Fluid response was de ned as a greater than 15% increase in SV after FC (21). This cutoff value was considered to be clinically relevant and in accordance with measurement variability. The non-parametric Wilcoxon rank sum test, Student's paired t test, Student's t test, and the Mann-Whitney test were used to assess statistical signi cance, as appropriate. A receiver-operating characteristic curve was established for the ability of E A , E es , the E A /E es ratio to predict a greater than 15% increase in SV. The limit for statistical signi cance was p < 0.05. SPSS® software (version 22, IBM, New York, NY, USA) was used for all statistical analyses.
Values for E A(ESP) , E es and E A /E es ratio were not signi cantly different between men and women, or according to type of surgery or medical characteristics (p value > 0.05), therefore allowing pooled analysis (Table 1). No patients developed complications (arrhythmia, hypoxaemia, left heart failure) during FC.   Fig. 2). In uid non-responders, FC was associated with higher values for CVP and E A(ESP) /E es ratio and a lower SW/PVA ratio ( Table 2).

Discussion
In uid responders, V-A coupling was characterized by a high pre-challenge E A /E es ratio (due to high E A ).
FC improvement in V-A coupling was associated to SV response. This effect was associated with a decrease in E A but no change in E es . Increased V-A coupling was associated with greater myocardial work e ciency. The pre-challenge E A /E es ratio was a good predictor of uid responsiveness.
Few studies have speci cally evaluated the effect of FC on V-A coupling. One study in cardiac surgery patients found an increase in SW, PVA and afterload, as a result of increased SV (10). A subsequent study of septic shock patients assessed the impact of FC on E A and its components, but did not measure E es (19). In contrast with the results mentioned above and the report by Mangano et al., we observed a decrease in arterial load after FC (10). Several explanations for our ndings can be proposed. The sympathetic nervous system plays a key role (via the barore ex) in regulating blood volume, blood ow and blood pressure (22). Accordingly, preload-dependent patients probably have higher levels of sympathetic activation than non-dependent patients, as evidenced by higher E A , higher SVRi, and lower C A values at a given blood pressure. This response is designed to adapt blood ow to the patient's needs, which appears to be effective, as ScVO 2 and arterial lactate levels were not signi cantly different between the two groups of patients. FC restores preload and CO, and thus meets the patient's needs. A decrease in E A might be caused by several interlinked mechanisms affecting the resistive component (HR and SVR) and the pulsatile component (C A ) of arterial load. The increase in blood pressure induced by an increase in CO decreases sympathetic activation, SVRi and E A . The barore ex has been shown to maintain adequate blood pressure by modulating E A , E es and blood volume (23). An increase in blood ow decreases vascular tone by activating the NO pathway and by initiating vascular recruitment (24). As a result of shear stress, blood ow modulates the diameter of blood vessels and can in uence aortic compliance (25). Segers et al. used a heart-artery interaction model to show that the contribution of the resistive component to this effect is threefold higher than that of the pulsatile component (26).
Our results evidenced a slight increase in LVEF in response to FC. This result is in line with the literature data; LVEF is described as a preload-dependent and afterload-dependent variable. Nevertheless, LVEF is more sensitive to afterload at low preloads (27). As LVEF can be expressed as E es /(E es + E A ), it is determined by V-A coupling (27). The observed increase in LVEF is therefore induced by a decrease in ventricular load, as represented by a decrease in E A and constant E es . These ndings are consistent with the results of studies of the effect of beta-blockers in patients with high blood pressure or heart failure (5).
In the present study, we demonstrated that the pre-challenge E A /E es ratio is also predictive of uid responsiveness. These results suggest that the differences between SV responders and SV nonresponders were mostly due to changes in V-A coupling. These results are not surprising and may be explained by the E A /E es ratio, which characterizes the interaction between the ventricle and the arterial vascular system. E es was assessed by the noninvasive method developed by Chen et al., in which measurements of pre-ejection and ejection times in patients are used to calculate E es (12,28). Studies have demonstrated that the aortic pre-ejection time can predict uid responsiveness (29). Cheng et al. subsequently demonstrated that the pre-ejection time/ejection time ratio was useful to identify heart failure patients with V-A uncoupling (30). Interesting, the cut-off found in the present study is close to the upper normal value of the E A /E es ratio.
This study presents a number of limitations. We speci cally included patients not treated with vasopressors and inotropes to avoid any treatment-related bias. Vasopressors are known to alter the cardiovascular response to FC (31). Hence, we can safely assume that our results were related to the sole effect of FC. The study population may have differed from septic shock patients. Most of our patients presented perioperative hypovolaemia, whereas septic patients generally have acute circulatory failure with a combination of hypovolaemia, changes in microvascular perfusion and central-to-peripheral arterial decoupling (32). The methods used to calculate E es and E A can be open to criticism because we did not use a high-delity ventricular pressure catheter. We measured E es by a noninvasive single-beat method based on a linear end-systolic pressure-volume relationship, and a constant volume axis intercept of the end-systolic pressure volume relationship (11,33). Calculation of E es assumes that the end-systolic pressure-volume relationship is load-independent, with a linear slope, and that V 0 is not in uenced by inotropes (33). We calculated ESP from a radial artery signal, which may differ from the aortic pressure signal (34). However, radial artery pressure has been reported to provide a good estimate of ESP (15,35).
We used several methods to assess E A that all gave similar results in the overall population and in the two groups, supporting the ndings of our study. Although it can be argued that estimation of ESP from the radial artery has not been fully validated, any error in this method would only affect the precision of absolute values of E A and E es, but not the E A /E es ratio, as the error in end-systolic pressure would be similar. The predictive value of E A /E es for increased SV can therefore be considered to be valid. Arterial load assessment was based on a two-element Windkessel model and integrative simpli cation. More precise models have been developed, such as three-and four-element Windkessel models that include arterial impedance and wave re ection. However, these methods would be di cult to apply at the bedside. Despite these limitations, noninvasive evaluation of E es and E A was validated against the gold standard method, and has been used in cardiac surgery (5)(6)(7). In the present study, E A and E es must be considered to be approximations of E A and E es . Study ow chart