Influence of Tidal Volume on Pulse Pressure Variations in Hypovolemic Ventilated Pigs with Acute Respiratory Distress-like Syndrome

• ❖ Inappropriate fluid administration to patients with acute respiratory syndrome can result in pulmonary and interstitial edema.

• ❖ Dynamic measures of fluid responsiveness have been questioned at small tidal volumes in critically ill patients suffering from acute respiratory distress syndrome.

• ❖ In healthy pigs, respiratory change in pulse pressure is a good indicator of volume status, but it is not a good indicator in pigs with acute respiratory distress syndrome and ventilated with small tidal volumes.

INAPPROPRIATE fluid administration to intensive care patients can result in pulmonary and interstitial edema.1,2Traditional markers of fluid responsiveness based on static preload measurements (i.e. , cardiac pressures and volumes) are not reliable predictors of fluid responsiveness because individual Frank-Starling curves may vary among patients, resulting in a positive response to fluid challenge (preload dependency) or no response (preload independency).3,4In contrast to static preload indices, dynamic indices are good indicators of volume and usually predict fairly well individual responses to volume loading.5–9Indeed, pulse pressure variation (ΔPP) was reported to accurately assess circulatory volume and predict fluid responsiveness in critically ill patients.3,4Other groups have used other indicators of stroke volume variations.5,6,10–15However, the majority of patients in these studies were ventilated with large tidal volumes (V^Tgreater than or equal to 8 ml/kg), and recent studies16–19have suggested that ΔPP is an unreliable indicator of fluid responsiveness at low V^Tin patients suffering from the adult respiratory distress syndrome (ARDS). This is particularly important because patients with acute lung injury or ARDS should be ventilated with V^Tless than 6 ml/kg.20,21 

Patients with ARDS have reduced lung compliance (C^L) because of stiffer lung parenchyma, and some authors have demonstrated that this phenomenon dampens airway pressure transmission to the pleural space.22On the other hand, a decrease in lung compliance contributes at the same time to an increase in transpulmonary pressure (P^tp; i.e. , pressure inside alveoli minus pleural pressure) for a given insufflated V^T. In this regard, we reasoned that to study the effects of a decreased V^Ton the accuracy of dynamic indexes in ARDS, we should also analyze effects on P^tpand esophageal pressure (P^es, pleural pressure) in this setting.17,23Moreover, to maintain minute volume ventilation in ARDS patients ventilated with low V^T, an increase in the respiratory rate (RR) is usually necessary.20This situation results in a lower number of heart beats being affected by each mechanical breath, a scenario that could also explain the low predictable value of ΔPP.24Indeed, Scharf et al.  25have already demonstrated that RR affects “reverse pulsus paradoxus.”

Because no study has evaluated the influence of V^Tand RR on ΔPP in mechanically ventilated animals with ARDS-like symptoms, the present study sought to determine the influence of these parameters on ΔPP value during hemorrhagic shock in mechanically ventilated pigs with normal lungs or after lung lavage-induced acute lung injury (ARDS-like syndrome).

Approval from the Ethics Committee for Animal Research of the University Medical Center and by the Cantonal Veterinary Office of Geneva, Switzerland, was achieved before the study was initiated. Handling of animals followed the guidelines laid out in the Guide for the Care and Use of Laboratory Animals (Institute of Laboratory Animal Resources, 1996). We studied 24 domestic pigs (weight ± SD, 31.4 ± 3.1 kg). Five pigs were used for pilot experiments, optimizing the study protocol, and three animals were excluded because of technical problems. These eight animals were not analyzed. Animals were assigned to groups sequentially; i.e. , we started with the control animals and after eight animals, we proceeded with the animals subjected to ARDS.

Animals had free access to food and water until 12 h before the beginning of experiments. Premedication consisted of intramuscular injections of 6 mg/kg azaperon, 0.5 mg/kg midazolam, and 0.5 mg atropine. Anesthesia was induced by isoflurane inhalation and maintained by 20 μg · kg⁻¹· h⁻¹fentanyl, 1.5–2.0% isoflurane, and 0.4 mg · kg⁻¹· h⁻¹pancuronium through a catheter placed into an ear vein. Depth of anesthesia was verified by paw pinch before muscular relaxation was administered. The animals were intubated and mechanically ventilated with oxygen in air (Fio²= 0.4 in control group [n = 8] and 1.0 in ARDS group [n = 8]) using a constant-volume respirator (Servo Ventilator 900; Siemens-Elema, Goeteborg, Sweden). No spontaneous breathing movements occurred during the experiments. V^Twas initially set at 10 ml/kg and the RR at 15 breaths/min (RR15) with a positive end-expiratory pressure (PEEP) of 0 cm H²O. Airway pressure and respiratory gases were continuously monitored (Ultima™; Datex/Instrumentarium, Helsinki, Finland). The right internal jugular vein, left internal carotid artery, and right femoral artery were cannulated for infusions, arterial pressure measurements, bleeding, and reperfusion. A flow-directed Swan-Ganz catheter (CCOmboV™, 7.5-French; Edwards Lifesciences, Irvine, CA) was introduced through the right internal jugular vein and advanced into the pulmonary artery to measure central venous pressure, mean pulmonary artery pressure, pulmonary capillary wedge pressure, continuous cardiac output, and continuous mixed venous saturation. Arterial pressure tracings and mean arterial pressure were measured by a left carotid arterial catheter advanced into the descending aorta. Vascular pressures were measured using calibrated pressure transducers (Honeywell, Zürich, Switzerland) positioned at the level of the left atrium. A standard 3-lead electrocardiogram was continuously displayed on a Hewlett-Packard monitor (M3150A; Hewlett-Packard, Andover, MA) with digital readout of heart rate by means of cutaneous electrodes. Diuresis was monitored by a suprapubic catheter.

Airway pressure was measured and V^Tcalculated by digital integration of a flow signal measured by a pneumotachograph (Godart 17212; Gould Electronics BV, Bilthoven, Netherlands), the two connected to the endotracheal tube. This setting permitted continuous breath by breath assessment of dynamic compliance of the respiratory system (C^RS= V^T/[plateau airway pressure − PEEP]) and expiratory airway resistance (R^aw=[plateau airway pressure − PEEP]/maximal expiratory air flow)26from three respiratory cycles. Measurement of P^eswas used to estimate pleural pressure and calculate P^tpby means of an inflatable latex double-balloon catheter (Ref C48 Sde Guenard; Marquat Génie Biomédical, Boissy-Saint-Léger, France) advanced into the esophagus with the distal balloon placed into the stomach and the proximal balloon in the esophagus behind the heart.27The two balloons were filled with air to a volume of 0.5 to 1 ml. Continuous P^tpwas calculated by subtracting simultaneous P^esfrom plateau airway pressure, and tidal P^tp(ΔP^tp) was determined as the amplitude of P^tpswings during each mechanical breath. Tidal P^es(ΔP^es) was also measured (i.e. , amplitude of P^esswings during a mechanical breath).

Systemic vascular resistance was calculated by dividing the difference between mean arterial pressure and central venous pressure measured at end-expiration by cardiac output, pulmonary vascular resistance by dividing the difference between mean pulmonary arterial pressure and pulmonary capillary wedge pressure by cardiac output. ΔPP was defined as the difference between the maximal and minimal values of pulse pressure divided by the average of the pulse pressure during a respiratory cycle.8 

Blood gas tensions, oxygen hemoglobin saturation and pH were intermittently analyzed by an automated oximeter (ABL-505 analyser; Radiometer, Copenhagen, Denmark). Arteriovenous oxygen content difference, intrapulmonary shunt, oxygen transport, oxygen extraction ratio, and oxygen consumption were calculated using standard formulae from the blood gas analysis and cardiac output measurement.

The different continuously recorded hemodynamic and respiratory measurements were stored at a sampling rate of 200 Hz via  an analog/digital interface converter (Biopac, Santa Barbara, CA) on a personal computer for off-line analysis. In all groups, lactated Ringer's solution was administered at 5 ml · kg⁻¹· h⁻¹to compensate for basal fluid loss. The animal's central body temperature was maintained at approximately 37.5°C with a warm air fan.

The different measured hemodynamic and respiratory variables were recorded during ventilation with a V^Tof 10 or 6 ml/kg and RR15 or RR25. Three variants of ventilation were successively used: V^T10 ml/kg at RR15, V^T6 ml/kg at RR25, and V^T6 ml/kg at RR15. Each ventilation mode was used during 5–7 min under control conditions during baseline, after blood removal (hemorrhagic shock of ∼30 min), and after reperfusion of the shed blood to restore normovolemia. The ratios of inspiratory time to expiratory time were not changed during the protocol. In the ARDS animal group, lung lavage-induced ARDS (∼ 1 h, see below) followed the baseline step and preceded hemorrhage and reperfusion.

After switching Fio²to 1.0, surfactant depletion was performed by repetitive lung lavage, instilling 1,000 ml NaCl, 0.9%, at 37°C into the trachea.28The procedure was repeated 3.1 ± 0.3 (SD) times at 12.6 ± 3.5-min intervals until criteria for ARDS-like syndrome were fulfilled (i.e. , Pao²/Fio²less than 200 mmHg). Fio²was kept at 1.0 throughout the remainder of the experiment in this group of pigs.

Forty percent of total blood volume was removed during 5–10 min via  the right femoral artery. Total blood volume was calculated as 75 ml/kg of body weight. The blood was stored in blood bags containing citrate-phosphate-dextrose (Baxter AG, Volketswil, Switzerland) and maintained at body temperature during continuous agitation until reperfusion. The time between the start of bleeding until completed reperfusion was 49 ± 5 min. The time between the start of bleeding and the start of reperfusion was 40 ± 4 min. The time between the end of bleeding and the start of reperfusion was 30 ± 4 min. The volume reperfused was 2.8 ± 0.3 ml · kg⁻¹· min⁻¹.

Group data are presented as mean values with 95% confidence interval (CI) in parenthesis, representing the continuous online recordings averaged over 30-s periods immediately preceding the intermittent recordings of pulmonary capillary wedge pressure and blood gas samples taken before the change to the next ventilation pattern. A general linear full factorial analysis of variance (ANOVA) model with repeated measures procedure was used to analyze the effects of the successive interventions (time as within-subjects factor using four levels, allowing determination of the effect of lung lavage, hemorrhage, and reperfusion on the various dependent variables), as well as the overall difference between the two groups of animals (treatment group as between-subject factor) using PASW Statistics 18 software package (SPSS, Inc., Chicago, IL). In each separated treatment group, the associated interaction between the within-subjects factor “time” and the between-subjects factor “ventilatory mode” was also analyzed to determine the effect of the ventilatory mode on the dependent variables. The two-sided Dunnett test for multiple comparisons with baseline values was used for post hoc  analysis when the ANOVA resulted in a P  value less than 0.05. Furthermore, a one-way ANOVA for repeated measures was used to detect statistical significance between the three ventilatory modes in a given condition, followed by the Bonferroni post hoc  test when the ANOVA resulted in a P  value less than 0.05.

Sixteen deeply sedated mechanically ventilated pigs were studied, eight in each group (control and ARDS). One set of recordings was lost in the control group during hemorrhage (animal 7; V^T6 ml/kg RR15 subgroup) and two sets in the ARDS group during control conditions (animal 11; V^T6 ml/kg RR15 and V^T6 ml/kg RR25 subgroups).

Lung alveolar lavage reproduced the expected hemodynamic, biologic, and respiratory characteristics of ARDS, with a 75% decrease in the Pao²/Fio²ratio (P < 0.001 between treatment groups throughout time after baseline time point, even if there was some recovery of this ratio with time; tables 1 and 2). ΔPP remained unchanged. ΔP^tpas well as ΔP^esincreased significantly after lung lavage (table 2).

After removal of 40% of total blood volume, there were marked expected changes in hemodynamic and respiratory variables (tables 1 and 2). Hemodynamics were similarly affected by hemorrhage in both control and ARDS pigs (table 1). However, compared with the control group and regarding lung mechanics, ΔP^tpdecreased significantly after hemorrhage in the ARDS group (two-way ANOVA, group effect during hemorrhage; P < 0.001, table 2), whereas ΔP^esremained unchanged (table 2). Retransfusion of the shed blood completely corrected hemodynamic variables in the control and ARDS groups, except for a persistent tachycardia (both groups) and a pulmonary vasoconstriction (ARDS group).

In the control group, lowering ventilation to a V^Tof 6 ml/kg reduced ΔPP by ∼40% regardless of circulatory state (P < 0.001 vs.  V^Tof 10 ml/kg). Furthermore, at each hemodynamic state, ΔPP values during both reduced V^Tmodes were significantly different from those through the 10 ml/kg ventilation (P < 0.01; table 3) regardless of respiratory rates. Despite reduced absolute ΔPP values observed during low V^Tventilation, their overall time course (time effect) was not significantly different from the 1 s recorded when ventilating with the larger V^T(time × ventilation interaction; P > 0.05), and their response to hemorrhage remained significantly different from respective baseline conditions at all RR values (P < 0.05; table 3; fig. 1, controls). In contrast, in pigs subjected to ARDS, lowering ventilation to a V^Tof 6 ml/kg rendered the ΔPP index unreliable for detecting hypovolemia regardless of the RR values (table 3; fig. 1, ARDS). However, when ΔPP was indexed and adjusted to ΔP^tp(fig. 1) and/or plateau pressure (Supplemental Digital Content 1, which is the figure displaying results, https://links.lww.com/ALN/A607), it became a reliable predictor for hypovolemia and reperfusion in ARDS animals ventilated with low V^T. A representative online recording of three consecutive periods of 20 s each at V^T10 ml/kg RR15, V^T6 ml/kg RR25, and V^T6 ml/kg RR15 is shown in figure 2.

The principal changes in pulmonary mechanics were as follows. First, significant changes in ΔP^tpin the control group after hemorrhage were absent, whereas a significant decrease in ΔP^tpwas observed in the ARDS group (table 3and fig. 3). Moreover, in this group, the hemorrhage-induced decrease in ΔP^tpvalues was proportionally larger (−22% [V^T6 ml/kg RR25] and −30% [V^T6 ml/kg RR15] compared with the decrease in pigs ventilated with a V^Tof 10 ml/kg (−12%; fig. 3, table 3). In the ARDS group ventilated with low V^T, P^esvalues decreased significantly during hemorrhage compared with baseline (P < 0.01). However, no significant changes were observed in ΔP^esafter hemorrhage regardless of V^Tor RR values (table 3).

The first finding of the present study was that after removal of 40% of total blood volume, ΔPP was a good indicator of hypovolemia in pigs with normal lungs and ventilated with low V^Tof 6 ml/kg RR15 and/or RR25, the level of this hypovolemia indicator being just reduced by approximately 40% compared with larger V^Tof 10 ml/kg. These results are in agreement with the cardiac output-preload dependency state anticipated by dynamic markers in critically ill patients with normal lungs and ventilated with low V^T.29–31However, our data highlight the result that ΔPP is not a sensitive marker of major changes in intravascular volume status in pigs affected by ARDS-like syndrome and ventilated with low V^Tregardless of RR, except when this indicator was adjusted to ΔP^tpand/or plateau pressure.18,19 

Because the slope of the Frank-Starling curve depends on ventricular contractility, static preload measurements are unable to anticipate the cardiac output-preload dependency.3In contrast, dynamic preload measurements, such as ΔPP determined by analysis of pulse pressure tracing, are useful in estimating the hemodynamic response to intravascular volume variations.3,4,32However, the magnitudes of dynamic preload indicators are also affected by V^Tduring acute lung injury.16,33Indeed, mechanical ventilation induces cyclic changes in intrathoracic and transpulmonary pressures, transiently affecting ventricular preload and leading to cyclic changes in stroke volume in preload-dependent but not in preload-independent patients. This phenomenon is related to the impact of inspiratory increase in intrathoracic pressure on pulmonary blood redistribution with changes in right ventricular outputs affecting left ventricular stroke volumes after two to three heart beats (pulmonary transit time).34According to this mechanism, inspiratory increase in left ventricular stroke volume results from delayed transmission through the pulmonary vasculature of the expiratory increase in right ventricular output and is followed by an expiratory decrease in left ventricular stroke volume as soon as the inspiratory decrease in right ventricular output has reached the left ventricle. If the decrease in right ventricular stroke volume after a mechanical breath is mainly related to decrease in venous return related to ΔP^esvalue in healthy lungs,35increase in ΔP^tpand right ventricular afterload could be an important determinant in lungs affected by ARDS.23 

In this regard, in the present study, we have analyzed the impact of ΔP^tpand ΔP^esin this setting.17,23An important finding of the present study is that ΔP^tpincreased after ARDS induction without significant changes in ΔP^es(tables 2 and 3). This fact highlights the idea that a decrease in C^RSin the ARDS group generates higher ΔP^tp. The major explanation for the inability of ΔPP to indicate hypovolemia in ARDS pigs ventilated with low V^Tcould be related to the increase in C^RSafter hemorrhage compared with control animals (table 2). This fact produced a decrease in a ΔPP determinant (ΔP^tp) in the ARDS group during hemorrhage when ΔP^tpvalues in the control group did not change (fig. 3and table 3). In addition, during hemorrhage, ΔP^tpin ARDS pigs ventilated with low V^Tvalues decreased more significantly, compared with the matched baseline values, than in ARDS pigs ventilated with V^Tof 10 ml/kg (fig. 3, table 3). The present explanation agrees with our results in that indexing ΔPP to ΔP^tpand/or plateau pressure made this marker a reliable predictor for hypovolemia and reperfusion (fig. 1and Supplemental Digital Content 1, https://links.lww.com/ALN/A607).18,19Because esophageal pressures are not usually available in patients, the present data highlight the value of ΔPP indexed to plateau pressure in this setting.19 

We could also expect that ventilation without PEEP in ARDS pigs could have led to a decrease in functional residual capacity with an increase in chest wall compliance, inducing a decrease in the pressure transmitted to the pleural space during mechanical breath. In the ARDS group, during hemorrhage, P^esvalues decreased more significantly during low VT ventilation compared with baseline (P < 0.01, table 3). This feature could have affected ventricular preload, cyclic changes in stroke volume, and ΔPP in pigs affected by ARDS and ventilated with low VT. However, our data (table 3) do not support this assumption because no significant changes were observed in ΔP^esvalues in pigs with healthy lungs and/or those affected by ARDS-like syndrome. Indeed, changes in venous return, stroke volume, and ΔPP are related to ΔP^esand not P^es.

In the present study, we have also reasoned that the cause of impaired relevance of this dynamic index in this situation could be related to the high RR associated with low V^Tin the ventilator setting of patients with ARDS and the impact of this fact on heart lung interactions.16,24With regard to this, we changed RR value during low V^Tas maintaining optimal volume minute ventilation in ARDS lungs requires increase in RR.20The rationale was that the heart rate/RR ratio and V^Tvalue are equally important as determinants of ΔPP.24Indeed, the number of heart beats affected by each mechanical breath is related to RR and inspiratory to expiratory time ratio. De Backer et al.  24found that high RR (30–40 breaths/min) could limit the predictive value of ΔPP for fluid responsiveness. We did not find any impact of RR on the reliability of ΔPP to detect hypovolemia in animals ventilated with low V^Tprobably for two reasons. First, compared with De Backer et al. ,24our highest level of RR was only 25 breaths/min in ARDS animals, which, in combination with tachycardia (table 1), made heart rate/RR unchanged in ARDS animals. Second, as stated previously, our model generated a change in respiratory mechanics that affected ΔP^tpas a determinant of ΔPP during hypovolemia.

The present animal study acknowledges some limitations. First, as discussed before, we used a low level of PEEP in the present animal study, whereas patients with ARDS ventilated with smaller V^Tare often treated with high levels of PEEP, which could have an impact on transpulmonary pressures and the pressures transmitted to pleural spaces. Second, our model of ARDS produces major changes in heart rate and respiratory mechanics, and we cannot exclude the idea that ΔPP would be insensitive to changes in volemia in the absence of major changes in these variables. Third, the internal carotid artery is closer to the central (aortic) pressure than the radial artery, and this fact could have affected ΔPP results. However, because the animals used in our experiments were healthy pigs, we expect that a peripheral artery would have been a less acceptable approach to measure pulse pressure, because the compliant aorta would have markedly buffered stroke volume.36Finally, the minor rise in diastolic pressure in pigs ventilated with low V^Tduring ARDS (fig. 3) suggests that a steady state was difficult to achieve and could have affected measurements.

In conclusion, we found ΔPP to be a reliable indicator of severe hypovolemia in pigs with healthy lungs regardless of V^Tor RR. In pigs affected by severe ARDS-like syndrome and ventilated with low PEEP, the measured index remains a valid indicator of severe hypovolemia at V^Tof 10 ml/kg but a less sensitive indicator at V^Tof 6 ml/kg. However, ΔPP indexed to ΔP^tpand/or plateau pressures were reliable gauges of blood volume status.

The authors thank Manuel Jorge-Costa and Sylvie Roule (Technicians, Anesthesia Investigations Unit, Geneva Medical School, Geneva, Switzerland) for excellent technical assistance.