Association Between End-Tidal Carbon Dioxide Pressure and Cardiac Output During Fluid Expansion in Operative Patients Depend on the Change of Oxygen Extraction

Abstract

In a model of hemorrhagic shock, end-tidal carbon dioxide tension (EtCO₂) has been shown to reflect the dependence of oxygen delivery (DO₂) and oxygen consumption (VO₂) at the onset of shock. The objectives of the present study were to determine whether variations in EtCO₂ during volume expansion (VE) are correlated with changes in oxygen extraction (O₂ER) and whether EtCO₂ has predictive value in this respect.

All patients undergoing cardiac surgery admitted to intensive care unit in whom the physician decided to perform VE were included. EtCO₂, cardiac output (CO), blood gas levels, and mean arterial pressure (MAP) were measured before and after VE with 500 mL of lactated Ringer solution. DO₂, VO₂, and O₂ER were calculated from the central arterial and venous blood gas parameters. EtCO₂ responders were defined as patients with more than a 4% increase in EtCO₂ after VE. A receiver-operating characteristic curve was established for EtCO₂, with a view to predicting a variation of more than 10% in O₂ER.

Twenty-two (43%) of the 51 included patients were EtCO₂ responders. In EtCO₂ nonresponders, VE increased MAP and CO. In EtCO₂ responders, VE increased MAP, CO, EtCO₂, and decreased O₂ER. Changes in EtCO₂ were correlated with changes in CO and O₂ER during VE (P < 0.05). The variation of EtCO₂ during VE predicted a decrease of over 10% in O₂ER (area under the curve [95% confidence interval]: 0.88 [0.77–0.96]; P < 0.0001).

During VE, an increase in EtCO₂ did not systematically reflect an increase in CO. Only patients with a high O₂ER (i.e., low ScvO₂ values) display an increase in EtCO₂. EtCO₂ changes during fluid challenge predict changes in O₂ER.

INTRODUCTION

The objective of volume-based hemodynamic resuscitation is to raise cardiac output (CO) and increase or restore the delivery of oxygen (DO₂) required to meet the demands of oxygen consumption (VO₂).¹ If DO₂ drops below a critical threshold, oxygen extraction (O₂ER) cannot increase in proportion to demand, and VO₂ becomes dependent on DO₂. Before critical O₂ER values arise, DO₂ can decrease independently of VO₂ (because DO₂ exceeds VO₂) and O₂ER will increase with demand (as demonstrated by a progressive fall in central venous saturation [ScvO₂]).² When O₂ER cannot rise any further, VO₂ decreases and the body's metabolism becomes partially anaerobic (with a resulting increase in blood lactate levels).³ In many pathological situations, VO₂ remains constant over a wide range of DO₂ values as a result of adjustments in tissue oxygen uptake.⁴ ScvO₂ is a clinically meaningful measure of tissue oxygenation,⁵ since it assesses the adequacy of DO₂ with regard to VO₂.⁶ Several studies have shown that ScvO₂-based hemodynamic resuscitation is associated with lower morbidity and mortality rates during anesthesia and intensive care.^7,8 Exhaled CO₂ (end-tidal carbon dioxide tension, EtCO₂) is also monitored in patients in the intensive care unit (ICU) or during anesthesia.⁹ Over short periods (and assuming a constant metabolic state), there is a qualitative relationship between EtCO₂ and CO.^10,11 Thus, EtCO₂ can be used as a noninvasive, continuous measure of CO during several clinical situations with low-flow states.^10–13 These results have not been confirmed in patients scheduled for surgery, in whom CO increased upon volume expansion (VE).¹⁴ One possible explanation is that patients scheduled for surgery and patients in the ICU differ in terms of systemic oxygen supply dependency. Most of the literature studies were performed in low-flow states, in which patients have been on a dependence phase between DO₂ and VO₂. This is probably not the case for most patients in the operating theatre. Based on a model of hemorrhagic shock in dogs, Guzman et al¹² demonstrated that EtCO₂ can reflect the dependence of DO₂ and VO₂ at the onset of shock and during hemodynamic resuscitation. Thereafter, Dubin et al¹³ confirmed the relationship between EtCO₂, DO₂, and VO₂. Lastly, EtCO₂ may be a noninvasive indicator of O₂ER (and its surrogate ScVO₂). Hence, the primary objective of the present study was to confirm that variations in EtCO₂ during VE are correlated with changes in O₂ER. We also evaluated the ability of variations in EtCO₂ to predict a decrease in O₂ER during VE.

METHODS

Ethics

The study's objectives and procedures were approved by the local independent ethics committee. Ethical approval for this study (Ethical Committee No. RNI2014-15) was provided by the Comité de Protection des Personnes Nord-Ouest II CHU—Place V. Pauchet, 80054 AMIENS Cedex 1 (Chairperson Bourgueil Thierry) on June 26, 2014. All patients received written information on the study and gave their verbal consent to participation prior to surgery. The present manuscript was drafted in compliance with the STROBE checklist for cohort studies.¹⁵

Patients

This prospective, observational study started on June 30, 2014 at Amiens University Hospital's cardio vascular and thoracic ICU over a 6-month period. The inclusion criteria were any major patient over 18 years, ventilated with controlled positive ventilation, for whom the physician decided to do a VE within hours of admission to the ICU. The indications for VE were arterial hypotension (systolic arterial pressure [SAP] lower than 90 mm Hg and/or mean arterial pressure [MAP] lower than 65 mm Hg), oliguria (urine output lower than 0.5 mL/kg per h over 1 h), clinical signs of hypoperfusion (skin mottling, capillary refill time over 2 s), and arterial hyperlactatemia (arterial lactate over 2 mmol/L). The noninclusion criteria were permanent arrhythmia, chronic obstructive pulmonary disease, and acute lung injury. The exclusion criteria were spontaneous ventilation, poor echogenicity, and arrhythmia.

Hemodynamic Parameters

An internal jugular vein central venous catheter and an arterial catheter were placed in all patients. Central venous pressure (CVP) and blood pressure were measured with a transducer zeroed at the mid-axillary line. Transthoracic echocardiography (Cx50, Philips Medical System, Suresnes, France) was performed by a physician who was blinded to the study outcomes. The left ventricular ejection fraction was measured using Simpson biplane method with a 4-chamber view. The diameter of the left ventricular outflow tract (LVOT) was measured on a long-axis parasternal view upon patient inclusion. Aortic area (SAo, in cm²) was calculated as π × LVOT²/4. The aortic velocity-time integral (VTIAo) was measured with pulsed Doppler and a 5-chamber apical view. Stroke volume (SV) (mL) was calculated as VTIAo × SAo. CO (in L/min) was calculated as SV × heart rate (HR). Mean echocardiographic parameters were calculated from 5 measurements (regardless of the respiratory cycle) and analyzed retrospectively. The intra and inter reproducibility of VTIAo measurements was tested prior to the study. Reproducibility values were 4.4 ± 3.9% and 4.4 ± 3.2%, respectively.

Oxygenation Parameters and EtCO₂

We recorded the ventilator settings (tidal volume, plateau pressure, and end-expiratory pressure) at baseline. Exhaled CO₂ was continuously measured at the tip of the endotracheal tube using a CO₂ cuvette with an integrated sensor (Drager, Luebeck, Germany). All parameters were measured on arterial and central venous blood gases. Arterial and venous blood gas levels, the lactate level, the blood hemoglobin concentration, and oxyhemoglobin saturation were assayed using an automated analyzer (ABL800 FLEX, Radiometer, Bronshoj, Denmark). Arterial oxygen content (CaO₂) and venous oxygen content (CvO₂) were calculated as follows: CaO₂ = 1.34 × Hb × SaO₂ + 0.003 × PaO₂; CvO₂ = 1.34 × Hb × ScvO₂ + 0.003 × PvO₂, where Hb is the hemoglobin concentration (in g/dL), PaO₂ is the arterial oxygen pressure (in mm Hg), SaO₂ is the arterial oxygen saturation (in %), PvO₂ is the venous oxygen pressure (mm Hg), ScvO₂ is the central venous oxygen saturation (in %), and 0.003 the solubility coefficient of oxygen. PCO₂ gap was calculated as follow: PCO₂ gap = PcvCO₂ − PaCO₂ (mm Hg).

DO₂ and VO₂ were calculated from arterial and central venous blood gases as follows: DO₂ (mL/min per kg) = (CaO₂ × 10 × CO)/weight; VO₂ (mL/min per kg) = the arteriovenous difference in oxygen content ([C(a − v)O₂] × CO × 10)/weight. O₂ER was defined as VO₂/DO₂ ratio. Arterial and venous CO₂ contents (CaCO₂, CvCO₂) were calculated according to Douglas Formula.¹⁶ The alveolar dead space (Vd/Vt) was estimated from EtCO₂ and PaCO₂ as (PaCO₂ − EtCO₂)/PaCO₂.¹⁷

Study Procedures

The following clinical parameters were recorded: age, gender, weight, and main diagnosis. First, a passive leg-raising (PLR) test was performed in order to evaluate the effects on SV, and assess preload status. After an equilibration period, baseline measurements of HR, SAP, MAP, diastolic arterial pressure (DAP), CVP, SV, CO, EtCO₂, and arterial/venous blood gas levels were obtained. In the present study, VE always consisted in infusing 500 mL of lactated Ringer solution over 10 min; 10 min after VE, a second set of measurements (SAP, MAP, DAP, HR, CVP, SV, CO, EtCO₂, and arterial/venous blood gas levels) was recorded. All patients had been sedated via the continuous infusion of propofol and were fully accustomed to mechanical ventilation. All patients underwent mechanical ventilation in volume-controlled mode with a tidal volume set to 7 to 9 mL/kg of ideal body weight, and a positive end-expiratory pressure of 5 to 8 cm H₂O. Ventilator settings (oxygen inspired fraction, tidal volume, respiratory rate, and end positive pressure) and norepinephrine dosage were not modified during the study period.

Statistical Analysis

We calculated that a sample of 50 patients would be sufficient to demonstrate a correlation of over 0.7 between EtCO₂, ERO₂, VO₂, DO₂, and CO. Fifty-five patients were therefore recruited, taking into account the exclusion criteria. The variables’ distribution was assessed using a Kolmogorov–Smirnov test. Data are expressed as the proportion (in %), the mean (standard deviation, SD) or the median (interquartile range), as appropriate. We measured the magnitude of EtCO₂ variations during VE by calculating the effect size (the mean divided by the SD).^18,19 The effect size was 0.74. Then, we calculated the coefficient of variation (CV), precision and least significant change (LSC) for EtCO₂. LSC is the least amount of EtCO₂ change that can be considered statistically significant; that is, the minimum percentage change between successive measurements that can be considered not due to random error and therefore representing a real change in ETCO₂. The EtCO₂ CV and LSC were determined in all studied patients at baseline during stable respiratory and hemodynamic conditions. The CV (95% confidence interval [CI]) was 1.8% (0.9–2.7) and the LSC (95% CI) was 2.5% (1.3–3.8). EtCO₂ responder was defined as an increase of EtCO₂ of more than 4% in EtCO₂ after VE. EtCO₂ nonresponder was defined as an increase of EtCO₂ of <4% in EtCO₂ after VE. This cut off correspond to LSC with its 95% CI.²⁰ Fluid responder was defined as an increase of more than 15% in the SV during VE.²¹ Fluid nonresponder was defined as an increase of <15% in the SV during VE. The nonparametric Wilcoxon rank sum test, Student paired t test, Student t test, and the Mann–Whitney test were used to assess statistical significance, as appropriate. Linear correlations were tested using Pearson or Spearman rank method. A receiver-operating characteristic curve was established for EtCO₂, with a view to predicting a decrease of over 10% in O₂ER, and a increase over 10% in ScVO₂.²² The threshold for statistical significance was set to P < 0.05. SPSS software (version 21, IBM, New York, NY) was used to perform statistical analysis.

RESULTS

Fifty-one postoperative patients were analyzed after inclusion in the study (Figure 1, Table 1). The indications for VE were as follows: arterial hypotension (n = 34), oliguria (n = 4), and clinical signs of hypoperfusion (n = 13). Eighteen (35%) patients had hyperlactatemia. Indications for VE did not differ between EtCO₂ responders and EtCO₂ nonresponders (P > 0.05). Twenty (39%) of the 51 patients were classified as EtCO₂ responders. All EtCO₂ responders were also fluid responders and displayed a mean (95% CI) EtCO₂ of 7% (6–9) during VE (Figure 1). Thirty-one patients were classified as EtCO₂ nonresponders and displayed a mean (95% CI) change in EtCO₂ of 0% (−1 to 1) during VE (Figure 1). Twenty-six of the EtCO₂ nonresponders were fluid responders and 5 were fluid nonresponders. At baseline, prevalence of norepinephrine treatment did not differ between the 2 groups of patients (12 [39%] EtCO₂ nonresponders vs 7 [35%] EtCO₂ responders, P = 0.1). At baseline, SV variations with PLR did not differ between EtCO₂ responders and EtCO₂ nonresponders (P > 0.05, Table 2).

FIGURE 1.

Study flow chart.

TABLE 1.

Characteristics of the Study Participants on Inclusion

TABLE 2.

Comparison of Hemodynamic Parameters in Nonresponders, EtCO₂ Nonresponders, and EtCO₂ Responders

Effect of VE on Hemodynamic and Blood Gas Parameters

In the study population as a whole, VE led to increases in MAP, CVP, SV, CO, EtCO₂, DO₂, PvO₂, and ScvO₂, and decreases in HR, PvCO₂, O₂ER, and alveolar dead space (Tables 2 and 3). At baseline, MAP, PvO₂, and ScvO₂ were lower and VO₂ and O₂ER were higher in EtCO₂ responders than in EtCO₂ nonresponders (regardless of the presence or absence of a fluid response in the latter; Tables 2 and 3).

TABLE 3.

Comparison of Blood Gas Parameters in EtCO₂ Nonresponders and EtCO₂ Responder Groups

In EtCO₂ nonresponders, VE led to increases in MAP, SV, CO, CVP, DO₂, and VO₂. PvCO₂ decreased during VE. In EtCO₂ responders, VE led to increases in MAP, SV, CO, CVP, EtCO₂, PvO₂, DO₂, and ScvO₂ and decreases in PvCO₂, PcCO₂ gap, O₂ER, and alveolar dead space (Tables 2 and 3).

Correlations Between Hemodynamic, Blood Gas Parameters, and EtCO₂

In the overall population at baseline, EtCO₂ was correlated with CO, DO₂, and O₂ER (r = 0.48, P = 0.001; r = 0.47, P = 0.001; and r = −0.42, P = 0.005, respectively). O₂ER was not correlated with arterial lactate levels (r = 0.01, P = 0.99 but was correlated with PavCO₂ (r = −0.46, P = 0.01). Changes in EtCO₂ during VE were correlated with those in CO, PvO₂, DO₂, VO₂, ScVO₂, and O₂ER (Table 4). A change of more than 4% in EtCO₂ during VE predicted a decrease of more than 10% in the VO₂/DO₂ ratio with an area under the curve (95% CI) of 0.88 (0.77–0.96) (P < 0.0001), a sensitivity of 71% (53–85), a specificity of 94% (71–100), a positive likelihood ratio of 12, negative likelihood ratio of 0.31, a positive predictive value of 96, and a negative predictive value of 62. In the same way, change in EtCO₂ during VE predicted an increase of more than 10% in ScVO₂ with an area under the curve (95% CI) of 0.90 (0.78–0.97), P < 0.0001.

TABLE 4.

Correlations Between Variations in EtCO₂, Hemodynamic Parameters, and Blood Gas Parameters During VE

DISCUSSION

Our results demonstrated that during VE, the occurrence of concomitant increases in EtCO₂ and CO depends on the relationship between VO₂ and DO₂; an increase in CO was not necessarily accompanied by an increase in EtCO₂. Only patients with a high extraction ratio (i.e., low ScvO₂ values) will display an increase in EtCO₂. Thus, changes in EtCO₂ during VE reflect changes in O₂ER in response to a rise in DO₂. Hence, during VE, EtCO₂ may be a useful noninvasive indicator of changes in systemic oxygen supply dependency when fixed ventilation is maintained.

Several preclinical and clinical studies have demonstrated that EtCO₂ can be used as a noninvasive, continuous measure of CO during low-flow states (cardiac arrest, hemorrhagic shock, cardiopulmonary resuscitation, circulatory shock, etc.).^9–11 Similarly, EtCO₂ has been shown to reflect changes in VO₂ and VCO₂ during hemorrhagic shock.^12,13 Our present results demonstrated that EtCO₂ may reflect changes in systemic oxygen supply associated with changes in CO during VE in nonseptic patients. All the patients in our study were postoperative sedated and nonhypothermic. Alveolar ventilation procedures and norepinephrine dosage did not change over the study period, and CO increased during VE. Even then, only 50% of fluid responders displayed an increase in EtCO₂, thus EtCO₂ was poorly correlated with changes in CO. Our results confirm previous findings in the operating theatre, where EtCO₂ and CO were rather low.¹⁴ To determine the mechanisms by which increase in CO increase EtCO₂ during VE, 1 would have to consider the study population and the effects of increase CO on blood gas parameters.

At baseline, EtCO₂ responders had a lower MAP than EtCO₂ nonresponders, whereas the 2 groups did not differ significantly in terms of preload status (i.e., variations in SV during PLR) and CO. Although EtCO₂ responders and EtCO₂ nonresponders did not differ in terms of DO₂, the EtCO₂ responders had a higher VO₂ and thus a higher O₂ER and lower ScVO₂. VE led to an increase in CO (and thus DO₂) and a decrease in PvCO₂ in fluid responders. Nevertheless, VE in EtCO₂ responders led to a recovery of DO₂ in consistent with oxygen needs: the decrease in O₂ER resulted in an increase in PvO₂ and ScvO₂. In contrast, EtCO₂ nonresponsiveness was associated with an increase in DO₂ and VO₂ because baseline O₂ER did not rise. Thus, concomitant increases in EtCO₂ and CO may result from several different mechanisms.

Under steady-state conditions, alveolar CO₂ elimination and therefore EtCO₂ depend on several factors: CO₂ production (VCO₂, due to metabolism), alveolar ventilation (mechanical ventilation), pulmonary perfusion (CO), and V/Q matching. VCO₂ depends on pulmonary elimination and metabolic production of CO₂. The changes in EtCO₂ in our population could not be explained by metabolic production of CO₂ for several reasons. In a model of hemorrhagic shock, Dubin et al¹³ demonstrated that VCO₂ could decrease EtCO₂ during the period of VO₂ supply dependency at low CO. The alterations in VCO₂ were statistically significant for changes in CO, DO₂, and VO₂ values that were greater than those observed in our study. A further mechanism might be related to removal of peripheral tissue CO₂ produced under anaerobic conditions.²³ In the present study, the baseline O₂ER values were below critical literature values at which tissue hypoxia was associated with anaerobic metabolism.²⁴ Moreover, no inter and intra group difference was shown for CaCO₂ and CvCO₂. One can hypothesize that decrease (increases) in O₂ER (ScVO₂) and CO will decrease the venous blood's capacity to carry CO₂ at a given PvO₂, which in turn will offset the increase in CO₂ delivery when CO rises.²⁵

Thus, an increase in DO₂ may decrease PvCO₂ and increase CO₂ delivery to the lung. At the same time, alveolar Vd/Vt fell in EtCO₂ responders (despite constant minute ventilation) as a result of 2 mechanisms. The increases in PvO₂ and CO may have improved alveolar perfusion pressure and the ventilation–perfusion ratio of the lung, which would tend to decrease PaCO₂.^26,27 In our population, changes in EtCO₂ had good correlation with changes in PvO₂ and ScVO₂ whereas they were not associated to those in PaCO₂ or PvCO₂. These mechanisms may explain (at least in part) why EtCO₂ did not change in EtCO₂ nonresponders, whereas CO did.

In summary, EtCO₂ responders had a low DO₂ with regard to their VO₂ resulting in higher O₂ER (lower ScVO₂). VE restored the relationship between VO₂/DO₂ through CO changes and increasing PvO₂ and CO₂ delivery to the lung, which improved the patients’ ventilation–perfusion ratio and thus increased EtCO₂.

The present study had a number of limitations. The study population (patients after heart surgery) may have differed from septic shock patients. Most of our patients suffered from acute circulatory failure as a result of perioperative hypovolemia, whereas septic patients generally have acute circulatory failure that combines hypovolemia, changes in microvascular perfusion and cellular dysoxia. A patient's response to VE, the relationship between DO₂ and VO₂, and the extent of anaerobic metabolism may depend on the etiology of acute circulatory failure.^28–30 We measured blood gas parameters from a central venous catheter and not from a pulmonary artery catheter. Although ScVO₂ cannot give a precise absolute estimate of SvO₂, it can serve as a guide to changes in SvO₂ and VO₂.^30,31 Monnet et al used blood gas parameters from a central venous catheter to assess VO₂, DO₂, and their changes over time during fluid expansion in septic patients.³¹ In our study, changes in ScVO₂ (measured) and O₂ER (calculated) had good correlation (r = −0.89, P < 0.0001). Moreover, predictive values of EtCO₂ changes during VE did not differ to predict an increase of ScVO₂ or a decrease of VO₂/DO₂ ratio. Since we performed repeated measurements of blood gas levels, mathematical coupling cannot be ruled out. But De Backer et al demonstrated that during controlled conditions, VO₂ calculated from hemodynamic data is a valid alternative to VO₂ derived from respiratory gas measurements.^30,32 The method used to calculate alveolar dead space was not standardized and may have introduced bias into the determination. The difference between EtCO₂ and PaCO₂ is altered in patients with altered ventilation/perfusion ratios (due to atelectasis, chronic heart failure, acute respiratory distress syndrome, etc.). In the present study, patients with chronic pulmonary disease or acute lung injury were excluded to limit this bias. Furthermore, these results cannot be extrapolated to EtCO₂ changes that result from the administration of vasopressor drugs.³³ EtCO₂ changes seem small but they are similar to those used to predict fluid responsiveness.³⁴ Lastly, given that we did not measure VCO₂, we cannot rule out the occurrence of changes in metabolic production of CO₂ during the VE-induced increase in CO.

CONCLUSIONS

During VE, an increase in CO was not necessarily accompanied by an increase in EtCO₂. Only patients with a high O₂ER (i.e., low ScvO₂ values) display an increase in EtCO₂. Thus, EtCO₂ changes during fluid challenge predict changes in O₂ER (i.e., ScVO₂) in response to an increase in DO₂. EtCO₂ may be a useful noninvasive indicator of changes in systemic oxygen supply dependency in operative patients when fixed ventilation is maintained.

UNCITED REFERENCES

¹⁸.