Acute hyperventilation increases the central venous-to-arterial PCO2 difference in stable septic shock patients

Background To evaluate the effects of acute hyperventilation on the central venous-to-arterial carbon dioxide tension difference (∆PCO2) in hemodynamically stable septic shock patients. Methods Eighteen mechanically ventilated septic shock patients were prospectively included in the study. We measured cardiac index (CI), ∆PCO2, oxygen consumption (VO2), central venous oxygen saturation (ScvO2), and blood gas parameters, before and 30 min after an increase in alveolar ventilation (increased respiratory rate by 10 breaths/min). Results Arterial pH increased significantly (from 7.35 ± 0.07 to 7.42 ± 0.09, p < 0.001) and arterial carbon dioxide tension decreased significantly (from 44.5 [41–48] to 34 [30–38] mmHg, p < 0.001) when respiratory rate was increased. A statistically significant increase in VO2 (from 93 [76–105] to 112 [95–134] mL/min/m2, p = 0.002) was observed in parallel with the increase in alveolar ventilation. While CI remained unchanged, acute hyperventilation led to a significant increase in ∆PCO2 (from 4.7 ± 1.0 to 7.0 ± 2.6 mmHg, p < 0.001) and a significant decrease in ScvO2 (from 73 ± 6 to 67 ± 8%, p < 0.001). A good correlation was found between changes in arterial pH and changes in VO2 (r = 0.67, p = 0.002). Interestingly, we found a strong association between the increase in VO2 and the increase in ∆PCO2 (r = 0.70, p = 0.001). Conclusions Acute hyperventilation provoked a significant increase in ∆PCO2, which was the result of a significant increase in VO2 induced by hyperventilation. The clinician should be aware of the effects of acute elevation of alveolar ventilation on ∆PCO2. Electronic supplementary material The online version of this article (doi:10.1186/s13613-017-0258-5) contains supplementary material, which is available to authorized users.


Background
Awareness that tissue hypoperfusion is a key factor in the pathogenesis of the multiple organ failures has focused the attention on surrogate indicators of tissue perfusion in the critically ill patient [1,2]. In this context, venous-to-arterial carbon dioxide tension difference (∆PCO 2 ) has been proposed as a marker of tissue hypoperfusion in patients with septic shock [3][4][5][6][7][8][9]. In fact, the increase in ∆PCO 2 that has been observed in low-flow states is mainly related to venous hypercapnia, rather than a decrease in arterial CO 2 partial pressure (PaCO 2 ), which can be explained by the tissue CO 2 stagnation phenomenon. Indeed, due to the decrease in transit time, there is a higher than usual addition of CO 2 per unit of blood passing the efferent microvessels, which leads to a rise in CO 2 partial pressure in the venous blood (PvCO 2 ) [10][11][12].
The reason for the preferred use of ∆PCO 2 over PvCO 2 as a marker of global tissue perfusion is that ∆PCO 2 was found to be less influenced by changes in PaCO 2 than PvCO 2 [13][14][15]. However, acute changes in PaCO 2 or arterial pH might have direct effects on microvascular tone and/or might induce variations in systemic oxygen consumption [15][16][17][18], and could then affect ∆PCO 2 . Recently, in healthy volunteers, acute hyperventilation was shown to be associated with an increase in the peripheral venous-to-arterial CO 2 difference due to a reduction in peripheral blood flow induced by acute hypocapnia [19,20]. Furthermore, Morel et al. [21] found, in a small study that included mechanically ventilated postoperative patients, that acute decreases in PaCO 2 resulted in significant increases in ∆PCO 2 without any change in cardiac output. Therefore, it is not clear so far what the impact of the rapid decrease in PaCO 2 on ∆PCO 2 is. This question is important because if PaCO 2 or arterial pH fluctuations could influence ∆PCO 2 , this effect will have to be taken into account by the physician when interpreting ∆PCO 2 at the bedside. The aim of this study was to investigate the impact of acute hyperventilation on ∆PCO 2 in mechanically ventilated and hemodynamically stable septic shock patients.

Methods
This prospective and observational study was conducted in a single general adult intensive care unit (ICU). The study was approved by our local institutional ethics committee (Comité d'éthique du Centre Hospitalier du Dr. Shaffner de Lens, France). Informed consent was obtained from each subject's next of kin.

Patients
The study included mechanically ventilated and hemodynamically stable septic shock patients. The diagnosis of septic shock was defined according to the criteria of the American College of Chest Physicians (ACCP)/Society of Critical Care Medicine (SCCM) Consensus Conference [22]. All patients had to be monitored by a transpulmonary thermodilution device (PiCCO, Pulsion Medical System, Munich, Germany) as part of routine management of septic shock in our ICU.
To avoid spontaneous breathing activity, patients remained sedated throughout the study via continuous infusions of propofol and remifentanil. Patients were ventilated in the control volume mode.
Exclusion criteria were pregnancy, age less than 18 years old, unstable hemodynamic condition (change in vasoactive drug dosage or fluid administration within 1 h preceding the protocol), high blood lactate levels (>2 mmol/L), and uncontrolled tachyarrhythmias (heart rate 140 beats/min).

Measurements
Demographic data, septic shock etiology, the Simplified Acute Physiology Score (SAPS) II, and the Sequential Organ Failure Assessment (SOFA) scores were obtained on the day of enrollment.
Cardiac index (CI) was obtained with the PiCCO monitor by triplicate central venous injections, in either the internal jugular or subclavian vein, of 20 mL of iced 0.9% saline solution and recorded as the average of the three measurements. In cases where the discrepancy in CI measurements was >10%, the measurement was repeated two more times (five times in total) with elimination of the highest and the lowest results.
Arterial and central venous blood gases analysis and arterial lactate levels were measured using the GEM Premier 4000 (Instrumentation Laboratory Co, Paris, France). To ensure accurate measurement, the blood gas analyzer was calibrated several times a day. Central venous blood was obtained from a central venous catheter with the tip confirmed to be in the superior vena cava at the entrance, or in the right atrium, by X-ray. ∆PCO 2 was calculated as the difference between central venous carbon dioxide tension (PcvCO 2 ) and arterial carbon dioxide tension (PaCO 2 ). Arterial oxygen content was calculated as CaO 2 (mL) = 1.34 × Hb (g/dL) × SaO 2 + 0.003 × PaO 2 (mmHg), where SaO 2 is the oxygen saturation of arterial blood, Hb the hemoglobin concentration, and PaO 2 the arterial oxygen tension. Central venous oxygen content was calculated as CcvO 2 (mL) = 1.34 × Hb (g/ dL) × ScvO 2 + 0.003 × PcvO 2 (mmHg), where PcvO 2 is the central venous oxygen tension and ScvO 2 the central venous oxygen saturation. DO 2 was calculated by using the formula: DO 2 (mL/min/m 2 ) = CaO 2 × CI × 10. VO 2 was calculated using the following formula: VO 2 (mL/ m 2 ) = CI × (CaO 2 − CcvO 2 ) × 10. Oxygen extraction was defined as: OE = VO 2 /DO 2 .
Heart rate (HR), mean arterial pressure (MAP), minute ventilation, respiratory rate, body temperature, and fractional inspired oxygen level were also recorded.

Study protocol
Patients were in steady state defined as less than 10% variation in HR, MAP, CI, and SaO 2 over a 60-min period before baseline measurements were initiated. Each patient was quiet and well adapted to the respirator. Fluid, doses of the vasopressor, and sedation drugs were kept constant in the hour preceding the measurements and throughout the study period. Variations in body temperature must have been <±0.5 °C. Enteral and/ or parenteral nutrition were continued and remained unchanged during the data collection period.
At baseline, a first set of measurements was performed, including hemodynamic and tissue oxygenation variables (HR, MAP, CI, VO 2 , ScvO 2 ), arterial lactate level, ∆PCO 2 , respiratory rate, and minute ventilation. Alveolar ventilation was then increased by raising the respiratory rate by 10 breaths/min (hyperventilation period). The inspiratory time was decreased to avoid the generation of an intrinsic positive end-expiratory pressure (PEEP) and to keep the level of plateau pressure constant throughout the study period. Also, the external PEEP remained unchanged. After 30 min of stabilization, a second set of measurement was recorded, including the same hemodynamic, respiratory, and tissue oxygenation variables (Additional file 1).
Changes in variables induced by the increase in alveolar ventilation were expressed as relative changes: [(variable after − variable before)/variable before] × 100.

Statistical analysis
Data are presented as mean ± SD or as median (25-75%, interquartile range). Normality was evaluated using the Shapiro-Wilk test. Comparisons of variables between before versus after increase in alveolar ventilation were assessed using Student's paired t test or Wilcoxon test, as appropriate. Linear correlations were tested using the Pearson or the Spearman test, as appropriate. The McNemar's test was used to compare two paired proportions.
In a previous study [23], we found that the smallest detectable difference for ∆PCO 2 was 2.0 mmHg. The smallest detectable difference is the minimum change (in absolute value) that needs to be measured by a laboratory analyzer in order to recognize a real change in measurement. Thus, for a power of 90% and α risk of 0.05, a sample size of 17 was required to detect a mean difference of 2.0 mmHg in ∆PCO 2 with a standard deviation of 2.35 mmHg [23].
Statistical analysis was performed using STATA 14.0 (StataCorp LP, College Station, Texas, USA). p < 0.05 was considered statistically significant. All reported p values are 2-sided.

Results
Eighteen septic shock patients were prospectively included in this study. Basic characteristics of the cohort are presented in Table 1. The principal source of infection was pneumonia (61%) with ICU mortality rate of 39%. All patients were sedated, mechanically ventilated without spontaneous breathing, and hemodynamically stable at their inclusions. No changes in vasopressor therapy and sedation level occurred during the observation period.

Effect of acute hyperventilation on blood gases, metabolic, and hemodynamic variables
Acute hyperventilation induced a significant increase in arterial pH and a significant decrease in PaCO 2 ( Table 2). Changes in arterial pH and PaCO 2 were paralleled by a statistically significant increase in VO 2 ( Table 2). Cardiac index, heart rate, and DO 2 remained unaffected.
Acute hyperventilation led to a significant increase in ∆PCO 2 , which resulted in a significant decrease in the number of patients with normal ∆PCO 2 value (∆PCO 2 ≤ 6 mmHg) ( Table 2). Furthermore, we observed a significant reduction in ScvO 2 in parallel to the increase in alveolar ventilation. Interestingly, lactate levels significantly increased when minute ventilation was increased ( Table 2). Abrupt elevation of alveolar ventilation was also associated with a significant increase in systemic vascular resistance index ( Table 2).

Correlation analysis
We found a good correlation between the increase in arterial pH induced by hyperventilation and the increase in VO 2 (r = 0.67, p = 0.002) (Fig. 1). Changes in PaCO 2 and VO 2 were also moderately correlated (r = −0.54, p = 0.02).
The increase in ∆PCO 2 observed after the elevation of alveolar ventilation was strongly associated with the increase in VO 2 (r = 0.70, p = 0.001) (Fig. 2a). Also, a strong relationship was found between the increase in VO 2 induced by hyperventilation and the decrease in ScvO 2 (r = −0.83, p < 0.001) (Fig. 2b). We found that the increase in lactate levels after the acute increase in alveolar ventilation significantly correlated with the increase in VO 2 (r = 0.54, p = 0.02).

Discussion
The main findings of our study were as follows: (1) acute increase in alveolar ventilation resulted in a significant increase in ∆PCO 2 accompanied with a significant decrease in ScvO 2 ; (2) these changes were linked to a significant increase in oxygen consumption induced by acute hyperventilation.
Early identification and treatment of tissue hypoperfusion are critical factors in the management of septic shock patients. In this regard, ∆PCO 2 has been considered as a marker that reflects the adequacy of tissue perfusion in septic shock states [3][4][5][6][7][8][9]. Increased ∆PCO 2 is associated with venous hypercapnia, which is explained by the low-flow-induced CO 2 stagnation phenomenon [11,12]. Venous hypercapnia results from insufficient elimination of the CO 2 produced by peripheral tissues, secondary to reductions in systemic and microcirculatory blood flow. However, under spontaneous breathing, hyperventilation may decrease PaCO 2 and thus may preclude the CO 2 stagnation-induced increase in PvCO 2 [24]. Because alveolar hyperventilation would decrease both arterial and venous PCO 2 without eliminating

Fig. 1 Relationship between changes in VO 2 and changes in pH induced by the increase in alveolar ventilation
the increased venous-to-arterial PCO 2 gap, it is recommended to assess ∆PCO 2 rather than only monitor PvCO 2 as a global marker of tissue perfusion [25].
However, a few studies have assessed the effects of acute hyperventilation on ∆PCO 2 in critically ill patients [13,14,21]. We found that the acute increase in alveolar ventilation led to a significant increase in ∆PCO 2 with an amplitude (2.2 mmHg) that was larger than its smallest detectable difference (2.0 mmHg) [23]. In addition, when the changes in ∆PCO 2 are expressed as relative changes, acute hyperventilation induced a significant increase in ∆PCO 2 with a magnitude (48.5%) that was also greater than its least significant change (32.4%) [23], which is the minimum change that needs to be measured by a laboratory analyzer in order to recognize a real change in measurement. In other words, the observed increase in ∆PCO 2 can be considered as a true change and was not due to a random variation. Our findings are in agreement with the results of Morel et al. [21]. Indeed, these authors studied the effects of an acute decrease in PaCO 2 , obtained by increasing the respiratory rate, on ∆PCO 2 in mechanically ventilated post-cardiac surgery patients. They found that acute hyperventilation provoked a significant increase in ∆PCO 2 (from 4.2 ± 1.8 to 7.6 ± 1.7 mmHg), while the cardiac index was unaffected. In that study [21], ScvO 2 also decreased in parallel with the increase in alveolar ventilation. Furthermore, in an animal study [16], the gradient between gastric mucosal PCO 2 and PaCO 2 (indicator of gut perfusion), obtained with gastric tonometry, increased significantly after hyperventilation. However, our results disagree with those of a previous study [13] that found no impact of hyperventilation on mixed venous-to-arterial PCO 2 difference in mechanically ventilated patients. In that study, the increase in alveolar ventilation was obtained very progressively by increasing the tidal volume from 7 to 10 mL/kg over a period of 3 h, which might explain the absence of changes in mixed venous-to-arterial PCO 2 difference. Also, the mean cardiac index at baseline was high (4.55 ± 0.90 mL/min/m 2 ), which would have prevented any increase in mixed venous-to-arterial PCO 2 difference by washing out any addition of CO 2 from the peripheral circulation.
Several mechanisms can be suggested to explain the increase in ∆PCO 2 observed in our study. A first potential explanation is that acute hyperventilation provoked the increase in systemic oxygen consumption and therefore CO 2 production. Thus, for a given venous blood flow, the increase in tissue CO 2 production should lessen the decrease in PcvCO 2 (induced by hyperventilation) relatively to the decrease in PaCO 2 , leading to a rise in ∆PCO 2 . We believe that such a mechanism may have contributed to the increase in ∆PCO 2 after acute hyperventilation in our study. Indeed, we observed a strong correlation between the increases in VO 2 between before and after hyperventilation and the increases in ∆PCO 2 (Fig. 2a). Also, the magnitude of the decrease in PcvCO 2 after hyperventilation was significantly less than the decrease in PaCO 2 (−16.5 ± 4.8 vs. −22.7 ± 5.5%, p < 0.001, respectively), explaining the observed increase in ∆PCO 2 . Similarly, the reduction in ScvO 2 found after hyperventilation can be explained by the increase in VO 2 . It is unlikely that the increase in VO 2 with hyperventilation was a result of an unstable state because of the lack of hemodynamic and temperature differences (Table 2), and the absence of changes in vasopressor and sedation drugs during the study period. We think that the observed increase in VO 2 was induced by acute hyperventilation since we found a good association between changes in pH and changes in VO 2 (Fig. 1). Acute respiratory alkalosis has been found, in some experiments in animals and humans, to increase VO 2 and CO 2 production independently of any significant hemodynamic changes [17,18,26,27]. Indeed, hyperventilation Fig. 2 Relationship between changes in VO 2 and the changes in ∆PCO 2 (a) and ScvO 2 (b) between after and before the increase in alveolar ventilation alkalosis, in mechanically ventilated dogs, increased VO 2 by 10-25% [17,18]. In anesthetized paralyzed patients, contradictory findings were observed with some authors reporting a significant increase in whole-body VO 2 [27], whereas others failed to demonstrate any significant variation [14]. Recently, Morel et al. [20], reported a twofold increase in VO 2 in healthy volunteers with hypocapnic condition compared to hypercapnic condition for the same minute volume, suggesting a possible contribution of this mechanism to the observed increase in peripheral venous-to-arterial CO 2 difference after induced acute respiratory alkalosis. The mechanism by which an acute respiratory alkalosis stimulates oxygen consumption is unclear and may involve many intracellular processes. A decrease in intracellular hydrogen ion concentration may stimulate the activity of phosphofructokinase, a key enzyme in the glycolytic cycle, which could result in increased intracellular adenosine triphosphate (ATP) hydrolysis and increased VO 2 [28,29]. Interestingly, we found a significant increase in lactate level induced by acute hyperventilation (Table 2). This finding could be an indirect marker supporting the activation of the phosphofructokinase enzyme and the increased rate of glycolysis in our study. Indeed, several studies reported increased lactate production with alkalosis [30,31], reflecting increased glycolysis.
A second possibility is that acute hypocapnia resulted in systemic vasoconstriction, thus decreasing the elimination of the total CO 2 produced by the peripheral tissues, and increasing the ∆PCO 2 . It has been shown that acute hypocapnia induces vasoconstrictive responses in various organs [14,32,33]. In healthy volunteers, Umeda et al. [19] observed that acute hyperventilation decreased both the minimal and mean flow velocity in the radial artery assessed by Doppler echography. The authors concluded that the decrease in mean blood flow, which was the result of increased vascular tone induced by hyperventilation, was responsible for the rise in peripheral venous-to-arterial CO 2 difference that they observed after acute hyperventilation. Similarly, Morel et al. [20] found a significant drop in the skin microcirculatory blood flow of healthy volunteers, evaluated with in vivo reflectance confocal microscopy, secondary to acute hypocapnia. In our study, we observed a significant increase in systemic vascular resistance in parallel with the elevation of alveolar ventilation (Table 2). Nevertheless, changes in systemic vascular resistance were not significantly correlated with changes in ∆PCO 2 nor with changes in ScvO 2 , which suggests, indirectly, a minimal participation of this mechanism to the increase in ∆PCO 2 . However, since we did not specifically evaluate the microcirculation we cannot eliminate or confirm the contribution of the vasoconstrictive mechanism to the observed increase in ∆PCO 2 secondary to acute hyperventilation.
A third possibility of the increase in ∆PCO 2 is that acute hyperventilation could induce variations in the PCO 2 /CO 2 content relationship. This mechanism is, however, unlikely to have occurred in our patients. Indeed, the relationship between CO 2 content and PCO 2 , which is curvilinear rather than linear, is influenced by many factors such as the degree of metabolic acidosis, the hematocrit, and the oxygen saturation (Haldane effect) [12,34]. Our patients did not have metabolic acidosis, and acute hyperventilation did not change the base excess meaningfully (Table 2). Although venous oxygen saturation decreased significantly after acute hyperventilation, it is unlikely that this change could have affected the PCO 2 /CO 2 content relationship, because first, it was not large; in this extent as stressed by Jakob et al. [35], changes in ∆PCO 2 might not parallel changes in CO 2 content differences under conditions of very low values of venous oxygen saturation (<30%), which was not the case in our patients. Second, if Haldane effect had affected the PCO 2 /CO 2 content relationship, it would have resulted in a decrease in ∆PCO 2 , rather than an increase in ∆PCO 2 [36].
Our results are of clinical importance. Indeed, changes in ventilator settings are regularly needed in mechanically ventilated patients. Since ∆PCO 2 is now widely recognized as a valuable marker to evaluate tissue perfusion in septic shock, a clinician should be aware that acute changes in pH or PaCO 2 induced by hyperventilation could impact ∆PCO 2 independently of changes in tissue perfusion. These findings should not dismiss the clinical value of ∆PCO 2 as a marker to detect tissue perfusion derangements. On the contrary, our results highlighted the usefulness of ∆PCO 2 , as an index of VCO 2 /cardiac output ratio, to detect the imbalance between the relative increase in VCO 2 and the blood flow, whatever the mechanism of this imbalance is (increases in oxygen consumption [37,38] or tissue hypoperfusion [9]).
We acknowledge some limitations to our study. First, the number of patients studied was small, but the study was sufficiently powered to detect a real change in ∆PCO 2 induced by hyperventilation. Second, the study was performed in a sample of septic shock patients from a single center, potentially limiting the generalizability of the results. However, our results confirm those of a previous study performed in a different patient population (post-cardiovascular surgery patients) [21]. Third, VO 2 was calculated from central venous oxygen saturation and not from mixed venous oxygen saturation or measured by indirect calorimetry, what might limit its accuracy. However, in our study, we were interested in investigating the changes in VO 2 induced by acute hyperventilation rather than by its absolute value. Furthermore, it has recently been demonstrated that calculating the oxygen-derived variables from the central venous blood allowed the detection of global tissue hypoxia in critically ill patients [39,40]. Finally, we did not evaluate the microcirculation, and thus, we were incapable of drawing any conclusions about the effects of acute hyperventilation on the local vascular tone and its relationship to ∆PCO 2 .

Conclusion
In stable septic shock patients, acute hyperventilation leads to a significant increase in ∆PCO 2 , which is associated with an increase in VO 2 , also induced by hyperventilation. Microcirculatory dysfunction is a key pathological mechanism involved in septic shock patients and could, also, be exacerbated by acute changes in alveolar ventilation. This finding should be taken into account when interpreting ∆PCO 2 at the bedside. Further studies are needed to investigate the effects of acute hyperventilation on systemic oxygen consumption and local vascular tone.