The effect of positive‐end‐expiratory pressure on stroke volume variation: An experimental study in dogs

Stroke volume variation (SVV) may be affected by ventilation settings. However, it is unclear whether positive‐end‐expiratory pressure (PEEP) affects SVV independently of the effect of driving pressure. We aimed to investigate the effect of driving pressure and PEEP on SVV under various preload conditions using beagle dogs as the animal model. We prepared three preload model, baseline, mild and moderate haemorrhage model. Mild and moderate haemorrhage models were created in nine anaesthetized, mechanically ventilated dogs by sequentially removing 10 mL/kg, and then an additional 10 mL/kg of blood, respectively. We measured cardiac output, stroke volume (SV), SVV, heart rate, central venous pressure, pulmonary capillary wedge pressure and the mean arterial pressure under varying ventilation settings. Peak inspiratory pressure (PIP) was incrementally increased by 4 cmH2O, from 9 cmH2O to 21 cmH2O, under PEEP values of 4, 8, and 12 cmH2O. The driving pressure did not significantly decrease SV under each preload condition and PEEP; however, significantly increased SVV. In contrast, the increased PEEP decreased SV and increased SVV under each preload condition and driving pressure, but these associations were not statistically significant. According to multiple regression analysis, an increase in PEEP and decrease in preload significantly decreased SV (P < .05). In addition, an increase in the driving pressure and decrease in preload significantly increased SVV (P < .05). Driving pressure had more influence than PEEP on SVV.


| INTRODUC TI ON
In humans, stroke volume variation (SVV) is a haemodynamic parameter that reflects fluid responsiveness. 1,2 SVV is derived from an arterial pulse contour analysis that reflects the respiratory changes in stroke volume (SV) under positive pressure ventilation. Several papers have reported that an increased SVV could be used as a sensitive indicator of fluid responsiveness. [1][2][3][4] However, past studies have reported that the following factors influenced SVV: tidal volume, compliance of thoracic wall, and positive-end expiratory pressure (PEEP). 5,6 Moreover, various respiratory settings are currently used in the care of critically ill patients; thus, we should understand the mechanism on how SVV changes under various respiratory settings and preloads. [7][8][9] Past studies reported that the changes in SVV directly reflected the fluid load in experimental dogs as well as humans. 10,11 Thus, we previously reported that the peak inspiratory pressure (PIP) significantly influenced SVV, and the association was enhanced by a decreased preload in ventilated dogs, because we investigate the effect of PIP and preload on SVV. 12 However, it is still controversial whether PEEP increases SVV or not. Past studies have reported that SVV increases according to the increase in PEEP because the SV is reduced. 13,14 However, these studies could not confirm the influence of PEEP on SVV, because PEEP and driving pressure changed simultaneously in these studies, with a fixed tidal volume under mechanical ventilation. 15 Thus, we aimed to investigate the effect of PEEP on SVV by adjusting various preload conditions and driving pressure, which is PIP minus PEEP, to differentiate the influence of PEEP and driving pressure with pressure-controlled mode under mechanical ventilations in animal experiment. 16

| Changes in the measured parameters in relation to PEEP and each preload
Dog body weights were 11.2 ± 1.0 kg. All haemodynamic data recorded during the study are presented in Table 1

| The relationship of driving pressure and PEEP and haemorrhage for SVV
In the multiple regression analysis, PEEP did not have an influence on SVV, but increasing driving pressure and decreasing preload significantly increased SVV. The regression coefficients of driving pressure and preload condition were 0.98 and 3.90, respectively ( Table 2). In addition, increasing PEEP and decreasing preload significantly decreased SV. The regression coefficients of PEEP and preload condition were −0.55 and −7.78, respectively (Table 3).

| D ISCUSS I ON
In this study, we demonstrated that PEEP decreased SV but did not increase SVV and driving pressure did not decrease SV but increased SVV under various preload conditions in experimental animals.
The strength of our study is that we investigated the conditions under fixed driving pressure but not under fixed tidal volume. We recognized that, under fixed tidal volume, increasing in PEEP elevates inspiratory pressure which correlates with SVV. 5,12 Therefore, a study about the effect of PEEP on SVV with a fixed tidal volume may mislead our interpretation on the association between PEEP and SVV. Understanding the influence of PEEP and driving pressure on SV and SVV can lead to optimal fluid management.
Stroke volume variation is calculated from maximal SV (SVmax) and minimal SV (SVmin). SVmax is a measure of the inspiratory elevation of the left ventricle SV under mechanical ventilation. 9,17,18 Additionally, the filling of the intrathoracic blood volume in the expiratory phase temporarily reduces LV preload and results in minimal SV (SVmin) at the beginning of expiration.
Driving pressure increase SVmax, and the difference between SVmax and SVmin increase. In fact, a previous study reported that a respiratory change of SV was low when a driving pressure was low. 19 On the other hand, PEEP reduces the intrathoracic blood volume resulting in reduction of both SVmax and SVmin, and SVV: the difference between SVmax and SVmin is not increased by increasing the PEEP. Thus, our results that PEEP did not significantly affect SVV despite the reduction of SV were reasonable and understandable.
Previous studies have reported that SVV and systolic pressure variation (SPV), which can be derived from arterial pressure curve by pulse counter analyses, increase according to the increase in PEEP. [13][14][15]20,21 Pizov 15 Similarly, Renner et al reported that SVV increases with increasing PEEP under mechanical ventilation with a fixed tidal volume of 10 mL/kg. 21 Given that their experiments were performed under volume-controlled ventilation, driving pressure was incrementally increased according to the increasing PEEP in healthy experimental animals. However, the major factor of elevating SVV should be driving pressure, not PEEP, because we reported that SVV correlates with driving pressure. 12 In their experiments, SVV may have been strongly influenced by driving pressure because they experimented with fixed tidal volume. Our experiment clarified that increasing driving pressure, rather than increasing PEEP, increases SVV.
On the other hand, Rose-Marieke et al reported that SV decreased but SVV did not increase according to the increase in PEEP, 22 with the driving pressure not changing along with increasing PEEP.
The discrepancy of these studies may be derived from unchanged driving pressure, while increasing PEEP.
There are several limitations. First, we did not investigate the level of severe haemorrhage and high-PEEP model to avoid mortality in the studied animals. The conditions of severe hypovolaemia and high-PEEP under mechanical ventilation may affect SVV. Second, ventilation settings used in healthy dogs cannot be extrapolated for humans. Lung compliance, vascular responsiveness, and pulse counter changes during tachycardia may differ between sick humans and healthy dogs. The lung compliance of dogs is so high that their tidal volume (TV) can reach >40 mL/kg for a maximum peak inspiratory pressure of 21 cmH 2 O. 12 We must perform a similar experiment because we cannot apply this experiment to humans as it is. Third, the number of experimental dogs was small. The sample size may be markedly small, and the results may be affected by various factors, individual dogs, and haemodynamics. We obtained accurate results by avoiding the unstable hemodynamic state; this was achieved by observing the animals for at least 2 minutes between steps to stabilize their hemodynamics, and we measured the parameters at least four times for every ventilation setting. Fourth, we administered TA B L E 1 Changes in the measured parameters in relation to PEEP and blood withdrawal In conclusion, we found that PEEP reduced SV but did not increase SVV under various preload conditions in the experimental animals. Driving pressure had more influence than PEEP on SVV.

| ME THODS
This study is an animal experiment using beagle dogs and was performed following the Science Council of Japan guidelines for animal experimentation. We obtained approval from the ethics committee for Animal Experimentation of Osaka Prefecture University, Japan.

| Experimental animals
Nine healthy beagle dogs, weighing 11.2 ± 1.0 kg, were used. The dogs used twice for the present study received with a minimum 21days period between experiments. The beagle dogs were purchased from Oriental Yeast Co. and bred in Osaka Prefecture University.
They were housed in separate cages, in which the temperature was maintained at 23 ± 1°Cand the light/dark cycle of time was 12 hours.
Feeding was once a day and water was available freely. All dogs were judged to be in good to excellent health based upon a physical examination, blood examination and chest radiography before each experiment by veterinarians. Food was withheld for at least 12 hours before drug administration, but the dogs were allowed free access to water prior to each experiment.

| Anaesthetic management
First, we subcutaneously injected 0.025 mg/kg atropine. We inserted a cannula into the peripheral vein and intravenously admin-

| Measurement of haemodynamic parameter under baseline model and haemorrhage models
We prepared the following three preload conditions: baseline model, re-infused with removed blood, which was temporarily stored in a blood bag during measurement. The dogs were followed up for a minimum 21-days period between experiments. All dogs were not killed, because a blood transfusion of 20 mL/kg at intervals more than 21-days is acceptable in the veterinary clinical practice. 25 There were no important adverse events or death in any dogs.

| Statistical analysis
All data were presented as the mean ± standard deviation (SD) or median (with interquartile range). We analyzed all data using the JMP Pro 12 software program for Windows (SAS Institute, Cary, NC, USA). Correlations between more than two variables were analyzed using a linear regression model based on the least-squares method. The univariate analysis was used to analyze the effect of the relationship between PEEP and haemorrhage, driving pressure, and haemorrhage on SVV. We entered haemorrhage, PEEP, and driving pressure as covariates and performed a multivariate regression analysis to understand the factor affecting SV and SVV. Differences were considered significant for P values <.05.

F I G U R E 3
The changes of ventilation settings under various preload conditions. We incrementally increased PIP by 4