Pulse-pressure variation and hemodynamic response in patients with elevated pulmonary artery pressure: a clinical study

A main goal of cardiovascular support is to restore and maintain sufficient cardiac output. Whereas prompt treatment of hypovolemia is necessary to sustain tissue perfusion, too much volume administration may result in edema formation and impaired tissue perfusion, with consequent organ dysfunction and increased risk of death [1‐5]. Because blood volume cannot reliably be assessed clinically [6‐11], the hemodynamic response to a volume challenge is commonly used to guide fluid management. Pulse-pressure variation (PPV), reflecting variations in stroke volume, has been advocated to discriminate between patients whose stroke volume increases in response to volume expansion and those who do not respond [12]. Conversely, PPV due to increased right ventricular afterload and right ventricular dysfunction may misleadingly suggest volume responsiveness [13‐15]. Increased pulmonary artery pressure and moderate or transient right ventricular dysfunction are not rare in intensive care patients requiring hemodynamic support. For example, transient right ventricular dysfunction occurs after cardiac surgery [16, 17], and high pulmonary artery pressures and dysfunction of both ventricles are common in septic shock [18, 19].

We hypothesized that PPV does not predict volume responsiveness in the presence of increased pulmonary artery pressure, and that right ventricular dysfunction may contribute to this. To test this hypothesis, we measured PPV and the hemodynamic response to clinically indicated fluid challenges in patients with septic shock and in postoperative cardiac surgery patients with increased pulmonary artery pressure.

The study was approved by the Ethics Committee of the Canton of Bern. Written informed consent was obtained from 20 cardiac surgery patients with a preoperative history of increased pulmonary artery pressure, and from both an independent physician and a close family member of 10 patients in septic shock.

Four cardiac surgery patients and two septic shock patients were excluded from the data analysis because of cardiac dysrhythmia during the study. Further, one cardiac surgery and one septic shock patient were excluded because of continuous spontaneous respiratory activity. Accordingly, 15 postoperative cardiac surgery patients and seven patients with septic shock were analyzed. Patient data are indicated in Table 1.

Sixty-nine fluid challenges were performed (44 in cardiac surgery patients and 25 in septic shock patients; one to eight per patient). The most common indications for fluid challenge were hypotension and peripheral vasoconstriction (Additional file 1).

Baseline hemodynamics (before the first fluid challenge) were similar in cardiac surgery and septic patients (Table 2). Overall, the fluid challenges increased stroke volume, CVP, pulmonary artery occlusion pressure (PAOP), and systemic and pulmonary arterial pressures (Table 3).

Nineteen (28%) of the 69 fluid challenges increased SV in 12 patients: in cardiac surgery patients, 12 of 44 fluid challenges (27%; seven patients) and in septic shock, seven of 25 fluid challenges (28%; five patients). If only fluid challenges that increased CVP by > 1 mm Hg are considered (31 of 44 fluid challenges in cardiac surgery and 22 of 25 in septic shock), SV increased in 15 (28%) of 53 fluid challenges.

The main finding of this study was the lack of association between PPV and volume responsiveness in patients with increased pulmonary artery pressure. This was the case both in postoperative cardiac surgery patients and in patients with septic shock, despite the very different underlying circulatory pathology. Enhanced pulse-pressure variation was present before most of the clinically indicated fluid challenges. Despite this, only about one of four fluid challenges resulted in increased stroke volume. The association was no better if only those fluid challenges were considered that were preceded by pulse-pressure variation of ≥13%, a threshold proposed by several authors, and approximately the mean value (12.5%) found in a recent systematic review [27].

Our results are in sharp contrast to those of several previous studies in septic and postoperative patients, in which pulmonary artery pressure was either not increased or not reported [6, 22‐26, 28‐33], or in which pulmonary artery pressure was markedly less increased [12, 34‐36]. Recently, Mahjoub et al. [15] reported a failure to predict fluid responsiveness by PPV in patients who had echocardiographic findings suggesting right ventricular systolic dysfunction without overt signs of right ventricular failure. We selected patients at relevant risk of acute right ventricular dysfunction due to increased pulmonary artery pressure. Our finding of reduced right ventricular ejection fraction in almost half of the nonresponders and in none of the responders supports the concept that right ventricular dysfunction contributed to the poor predictive value of PPV. Admittedly, the reliability of right ventricular ejection fraction estimation by using thermodilution catheters is controversial. Because we did not perform concomitant echocardiography, it is conceivable, conversely, that the presence and severity of right ventricular dysfunction might have been underestimated.

Failure to induce an increase in preload and failure to detect changes in stroke volume would be the most obvious alternative explanations for this controversial finding. We consider these explanations unlikely. First, any acute increase in central venous pressure in response to volume loading suggests an increase in preload and should result in an increase in stroke volume if the heart is operating in the volume-responsive part of the cardiac-function curve [20, 21, 37]. The increase in central venous pressure in 76% of fluid challenges strongly suggests that preload acutely increased in the majority of fluid challenges; in those without a relevant increase in central venous pressure, the fluid challenge apparently failed to increase the stressed volume, possibly due to vasodilation. Second, the percentage of responders versus nonresponders (27% versus 73%) was the same when all fluid challenges were considered. Because no clinically applicable method provides an accurate beat-to-beat measurement of cardiac output, we chose to use the continuous thermodilution method. We used the 1-minute values provided by the software, so this method does not induce a clinically relevant delay in the measured values. Hence, relevant rapid changes should not have been missed. We also eliminated all phases of spontaneous respiration by evaluating all individual tracings, and used tidal volumes large enough (9 ml/kg on average (range, 8 to 10 ml/kg)) to result in relevant intrathoracic pressure changes [36, 37].

The potential of increased pulmonary artery pressure to interfere with the predictive value of pulse-pressure variation has been addressed previously, but its clinical relevance has been considered small [13, 14]. This view is in sharp contrast to the frequent occurrence of high pulmonary artery pressures in the early postoperative period after cardiac surgery [16, 17], as well as in septic shock [18, 19]. Because the use of the pulmonary artery catheter has decreased, moderate or transient but relevant increases in pulmonary artery pressures may easily be overlooked. Our results suggest that under these conditions, PPV may erroneously suggest volume responsiveness. The recent study of Mahjoub et al. [15] supports this concept. Those authors considered the risk of false-positive prediction of fluid responsiveness by PPV high enough to warrant echocardiography before performing fluid challenge based on increased PPV.

The dynamic association between volume status and PPV has been advocated as an alternative to predict which patients will respond to fluid administration [12, 14]. We tested this in a clinical context in which fluid infusion is unavoidable (that is, in patients with septic shock and in patients immediately after cardiac surgery). Instead of selecting patients with a specific condition believed to reflect fluid responsiveness, we chose situations in which clinicians considered volume expansion with fluids necessary. Although PPV has been well validated in the context of hypovolemia [38, 39], the presence of pure hypovolemia is probably less common and more difficult to treat in patients with septic shock or after cardiac surgery, when pulmonary artery pressures are increased. In these situations, complex heart-lung interactions may be present as a consequence of either compromised myocardial function in cardiac surgery patients or both diastolic and systolic biventricular dysfunction and acute changes in pulmonary vascular resistances in patients with septic shock. Previous studies in cardiac surgery patients in whom PPV predicted volume responsiveness either have been performed right after induction of anesthesia [23, 24, 28‐30], have included patients with normal pulmonary artery pressure [24, 26, 30, 31], or have not reported pulmonary artery pressure [22, 23, 28, 29, 32]. In studies on sepsis patients, pulmonary artery pressure was less increased than in the present study [12, 34‐36] (average mean pulmonary artery pressure, 24 to 26 mm Hg before fluid challenge versus 33 mm Hg in this study) or was not reported [25, 33].

We believe that our findings are real, especially considering the persisting large PPV values. These increased PPV values cannot be explained by relative hypovolemia, but are rather the consequence of an impeded right systolic function, as a consequence of intrathoracic pressure changes in a situation in which the right ventricle is already at the flat part of preload dependence. Hence, PPV had no predictive value for volume responsiveness. Furthermore, clinical use of PPV to predict volume responsiveness may misleadingly suggest volume responsiveness when increased pulmonary artery pressure compromises right ventricular function, and further volume expansion may be harmful.

We therefore performed a second, experimental study. In the accompanying article [40], we assess the effect of acutely increased pulmonary artery pressure on volume-responsiveness prediction with PPV.

A limitation of this study is the lack of airway-pressure signal recording concomitant with the hemodynamic signals. Our approach does not guarantee 100% exclusion of data from patients with spontaneous respiratory activity, although respiratory efforts with an influence on PPV should mostly be detectable in central venous and pulmonary artery pressure tracings. Recording airway-pressure signals would also have enabled us to define whether PPV was the result of an increase in PP during inspiration, a decrease during expiration, or both.

Another potential confounder is the use of a colloid rather than a crystalloid. It has been suggested that colloids may have effects on cardiac function independent of their volume-expanding effect. However, this should have tended to increase fluid responsiveness rather than to decrease it.

We conclude that both early after cardiac surgery and in septic shock, patients with increased pulmonary artery pressure respond poorly to fluid administration. Under these conditions, PPV cannot be used to predict fluid responsiveness. We suggest that right ventricular dysfunction contributed to the poor response to fluids.