Arterial to end-tidal carbon dioxide tension difference in children with congenital heart disease{dagger}

J. A. Short1,5, S. T. Paris2, P. D. Booker3 and R. Fletcher4

1Royal Liverpool Children’s Hospital, Eaton Road, Liverpool L12 2AP, UK. 2St Luke’s University Hospital, Gwardamangia, Malta. 3University of Liverpool, Liverpool, UK. 4Manchester Royal Infirmary, Manchester, UK 5Present address: Royal Hallamshire Hospital, Sheffield, UK*Corresponding author

{dagger}Presented at the meeting of the Anaesthetic Research Society, Liverpool, March 2000.

Accepted for publication: October 17, 2000


    Abstract
 Top
 Abstract
 Introduction
 Theory
 Methods
 Results
 Discussion
 References
 
In children with congenital cyanotic heart disease, right-to-left intracardiac shunting causes an obligatory difference between arterial and end-tidal carbon dioxide tension (PaCO2PE'CO2) as venous blood, rich in carbon dioxide, is added to the arterial circulation. This obligatory PaCO2PE'CO2 difference, which can be predicted from knowledge of oxygen saturation, haemoglobin concentration and PaCO2, increases as oxygen saturation decreases, most markedly when the haemoglobin concentration is high. A second possible cause of the PaCO2PE'CO2 difference is the effect of pulmonary hypoperfusion caused by the shunt. We studied 60 children undergoing cardiac surgery and compared the predicted the PaCO2PE'CO2 difference with measured values to investigate the extent to which additional factors influence the clinically observed PaCO2PE'CO2. In many children, observed values were much greater than predicted, which is compatible with some degree of pulmonary hypoperfusion. However, this was not felt to represent the complete picture in all patients. Another cause of ventilation–perfusion mismatch was suspected in those children who showed a considerable improvement in oxygen saturation during ventilation with an increased FIO2. We believe that pulmonary congestion caused by large left-to-right shunts may further increase the PaCO2PE'CO2 difference.

Br J Anaesth 2001; 86: 349–53

Keywords: carbon dioxide, elimination; heart, congenital defects; blood, haemoglobin


    Introduction
 Top
 Abstract
 Introduction
 Theory
 Methods
 Results
 Discussion
 References
 
In children with normal cardiorespiratory function, arterial carbon dioxide tension (PaCO2) during anaesthesia approximates to end-tidal carbon dioxide tension (PE'CO2).1 2 In cyanotic children, however, the right-to-left intracardiac shunt has important effects on gas exchange. The addition of venous blood, relatively poor in oxygen and rich in carbon dioxide, to the systemic ventricle of the heart causes not only a reduction in arterial oxygen saturation, but also elevation of arterial PCO2 above pulmonary capillary, alveolar, and end-tidal PCO2. This reduction in the efficiency of carbon dioxide elimination, an apparent alveolar deadspace, is a primary effect of right-to-left shunting. Right-to-left shunting also reduces pulmonary blood flow, potentially causing alveolar hypoperfusion. This secondary effect increases alveolar deadspace still further, and has an additive effect on the arterial–alveolar PCO2 difference and thus the PaCO2PE'CO2.

The theoretical relationship between the PaCO2PE'CO2 difference and the arterial saturation in patients with right-to-left shunting has been described by Fletcher.3 As saturation falls, the PaCO2PE'CO2 difference increases proportionately. The increase in the PaCO2PE'CO2 difference with decreasing saturation is greatest when the haemoglobin concentration is high. This relationship was confirmed in children with pulse oximeter saturations of 60–100%.3

Thus, in the absence of pulmonary hypoperfusion, PaCO2 can be predicted accurately in cyanotic children with otherwise normal lung function from knowledge of the PE'CO2, haemoglobin concentration and saturation.4 However, should significant pulmonary hypoperfusion exist, then the PaCO2PE'CO2 difference will always be greater than its estimate from first principles. The difference increases as saturation decreases, unless there is a compensatory increase in cardiac output, as may occur in severely hypoxaemic children.4

We investigated whether the effect of pulmonary hypoperfusion could be detected by comparing observed and predicted values for PaCO2PE'CO2 in children with congenital heart disease.


    Theory
 Top
 Abstract
 Introduction
 Theory
 Methods
 Results
 Discussion
 References
 
The theoretical relationship between the PaCO2PE'CO2 difference and saturation can be defined by the following equation:3

(PaCO2PE'CO2)=RxHbx0.0131x(100–SaO2)/k

where R is the respiratory quotient, Hb is the haemoglobin concentration in grams per litre, 1.31 is the combining constant of haemoglobin in millilitres of oxygen per gram of haemoglobin, and k is the slope of the carbon dioxide dissociation curve for whole blood. It can be seen that PaCO2PE'CO2 will increase as saturation decreases and haemoglobin increases, reflecting the primary effect of the right-to-left shunt. This predicted difference takes no account of the effect of pulmonary hypoperfusion.

Calculation of predicted (obligatory) PaCO2PE'CO2 difference in patients with intracardiac shunting
A. Calculate total arterial carbon dioxide content
(i) Using measured PaCO2, determine bicarbonate carbon dioxide (ml litre–1) from Fig. 1.



View larger version (41K):
[in this window]
[in a new window]
 
Fig 1 Components of the carbon dioxide dissociation curve for whole blood. Dissolved carbon dioxide and bicarbonate ion vary with PCO2 but are little affected by the state of oxygenation of the haemoglobin. Carbamino carriage of carbon dioxide is strongly influenced by the state of oxygenation of haemoglobin but hardly at all by PCO2. (Reproduced with permission from Nunn’s Applied Respiratory Physiology.5)

 
(ii) Using measured PaCO2, Hb and saturation, determine carbamino carbon dioxide (ml litre–1) from Fig. 2 (1 mmol litre–1=22.4 ml litre–1).



View larger version (25K):
[in this window]
[in a new window]
 
Fig 2 The broken lines on the graph indicate the carbamino carriage of carbon dioxide at different levels of saturation of haemoglobin with oxygen (Hb 15 g dl–1). A represents arterial blood and V represents mixed venous blood in patients with no gas exchange lesion. (Reproduced with permission from Nunn’s Applied Respiratory Physiology.6)

 
(iii) Add (i) to (ii) above to give total arterial carbon dioxide content.

B. Calculate pulmonary end-capillary carbamino content
Assume 100% saturation and standard carbamino content at haemoglobin concentration=15 g dl–1 of 22 ml litre–1 carbon dioxide. Correct for measured haemoglobin concentration.

C. Calculate arterial minus end-capillary oxygen difference
(i) End-capillary oxygen content=Hbx1.31+[((FIO2x (PB–PH2O))–(PE'CO2/R))x0.225] [using Hb in g litre–1, PE'CO2 to estimate alveolar PCO2 (kPa) and assuming standard values for barometric pressure (PB), PH2O and R (0.8)].

(ii) Arterial oxygen content=(Hbx1.31xSaO2/100)+(PaO2x0.225) (using Hb in g litre–1, PaO2 in kPa).

(iii) Subtract (i) from (ii) above for arterial minus end-capillary oxygen difference (ml litre–1).

D. Calculate arterial–end-capillary carbon dioxide difference
Multiply C (iii) above by R (assume standard value 0.8 for respiratory quotient).

E. Calculate pulmonary end-capillary bicarbonate carriage
End-capillary bicarbonate=total arterial carbon dioxide content [A (iii)] – carbon dioxide difference (D) – end-capillary carbamino content (B) (ml litre–1).

F. Determine end-capillary PCO2
Using the derived bicarbonate value from E above, determine the end-capillary PCO2 from Fig. 1. This approximates to the predicted end-tidal PCO2 value and can be subtracted from the measured PaCO2 to determine the predicted PaCO2PE'CO2.


    Methods
 Top
 Abstract
 Introduction
 Theory
 Methods
 Results
 Discussion
 References
 
After local Ethics Committee approval and written informed consent had been obtained from the parents, 60 children aged between 3 days and 10 yr who were undergoing cardiac surgery were entered into the study. Children with non-cyanotic congenital heart disease and profoundly cyanotic children were included to give a wide range of saturation levels and degrees of shunt. The oxygen saturation in air, measured by pulse oximetry, was recorded for each child before anaesthesia was induced.

Anaesthesia was induced with thiopentone 4 mg kg–1 and neuromuscular block obtained with vecuronium 0.1 mg kg–1 followed by 0.2 mg kg–1 h–1. Anaesthesia was maintained with 0.5–1% isoflurane in oxygen and/or air, supplemented with i.v. alfentanil 10 µg kg–1, followed by an infusion at 2 µg kg–1 min–1. All patients were ventilated in a standardized fashion, using a Servo 900C ventilator (Siemens, Bracknell, UK), set to deliver 10 ml kg–1 in volume control mode at a rate of 15–25 bpm. The inspiratory time totalled 40% of the respiratory cycle, including a 10% end-inspiratory pause. An end-expiratory pressure of approximately 2 cm H2O was applied routinely. Intrinsic positive end-expiratory pressure was not measured. During chest opening by thoracotomy or sternotomy, we made simultaneous recordings of oesophageal temperature, haemoglobin concentration and oxygen saturation, oxygen and carbon dioxide tensions in arterial blood, and inspired oxygen and expired carbon dioxide concentrations.

Blood gas and co-oximetry analysis was carried out using a Chiron Diagnostics 865 blood gas machine (Bayer, Newbury, UK). The anaesthetic gases were measured with a Capnomac Ultima gas monitor (Datex Engstrom, Sidcup, UK), which was calibrated against the blood gas machine each morning using a standard gas calibration mixture. The procedure involved initial analysis of the calibration gas by the blood gas machine, and noting the PCO2 measured by the machine. The capnograph was then calibrated using the calibration gas in the usual way, but entering the PCO2 concentration as measured by the blood gas machine rather than that printed on the gas canister.

For endotracheal tube sizes 3.0–4.5 mm, gas was sampled for inspired oxygen concentration and end-tidal carbon dioxide concentration using a special connector that allowed gas collection from the proximal endotracheal tube. This allowed greater accuracy of measurement in small subjects.2 In older children with larger endotracheal tube sizes, normal sidestream gas sampling was used. All blood gas samples were temperature-corrected at analysis.

The data collected were then used to calculate predicted values for PaCO2PE'CO2 as described previously. Carbon dioxide dissociation curves (Figs 1 and 2) were reproduced from Nunn’s Applied Respiratory Physiology5 6 and were enlarged and annotated with grid-lines to aid accuracy. The observed values recorded for arterial and end-tidal carbon dioxide tensions were compared with the predicted values by the use of paired t-tests and regression analysis.


    Results
 Top
 Abstract
 Introduction
 Theory
 Methods
 Results
 Discussion
 References
 
Complete data were obtained from 60 children with ages ranging from 3 days to 10 yr (median 10 months) and saturations in air ranging from 65 to 97% (median 85%). Diagnoses included transposition of the great arteries (nine patients), tetralogy of Fallot (22 patients), hypoplastic left heart syndrome (11 patients), tricuspid atresia (one patient), pulmonary atresia (nine patients), atrioventricular canal defects (two patients), ventricular septal defects (four patients), interrupted aortic arch (one patient) and anomalous pulmonary venous drainage (one patient).

The predicted PaCO2PE'CO2 value was calculated for each patient as described above. It was found that five patients within the first 10 had physiologically inconsistent data. One had a very negative PaCO2PE'CO2 and four had a normal PaCO2PE'CO2 with oxygen saturation less than 85%, a situation incompatible with the shunt equation. It was decided to remove these patients from the study, but, in order to reduce bias, data obtained from all of the first 10 patients were eliminated from further analysis.

The observed and predicted values for PaCO2PE'CO2 for the remaining 50 patients were plotted against arterial saturation (measured by the co-oximeter), as shown in Fig. 3. The mean observed PaCO2PE'CO2 difference was 1.07 (SD 0.58) kPa and the mean predicted value was 0.42 (0.29) kPa (P<0.0001). The figure shows that many of the observed values for PaCO2PE'CO2 were considerably greater than predicted. Furthermore, the observed PaCO2PE'CO2 exceeded zero in many of the well-oxygenated patients. In fact, the regression line indicated a difference of 0.7 kPa at 100% saturation. The expected negative correlation between PaCO2PE'CO2 and oxygen saturation was apparent for the predicted data; the correlation was less strong for the observed data.



View larger version (25K):
[in this window]
[in a new window]
 
Fig 3 The observed and predicted PaCO2PE'CO2 differences, plotted against arterial saturation measured by pulse oximetry, for 50 children undergoing cardiac surgery. The solid line represents the regression line for predicted data [regression equation: difference=3.45–0.033(SaO2), r=–0.92, P<0.0001] and the broken line represents the regression line for the observed data [difference=4.5–0.038(SaO2), r=–0.52, P=0.0001]. The observed values are shown in two groups to indicate which patients had clinical evidence of pulmonary congestion before surgery.

 

    Discussion
 Top
 Abstract
 Introduction
 Theory
 Methods
 Results
 Discussion
 References
 
The results of this study suggest that the PaCO2PE'CO2 differences in these 50 children with congenital heart disease were frequently greater than that caused solely by right-to-left shunting. The values of PaCO2PE'CO2 predicted from the primary shunt effect show the expected increase in PaCO2PE'CO2 of 0.3–0.4 kPa for every 10% fall in saturation.3 Clearly, there are other factors contributing to the derangement in gas exchange. In patients with pure right-to-left shunts, these might include the secondary effect of pulmonary hypoperfusion. However, the results are further complicated by the fact that even the well- oxygenated children in our study did not appear to have normal carbon dioxide elimination, in contrast to the results of Fletcher et al.7 In that study, children with normal oxygen saturation were found to have a zero or negligible PaCO2PE'CO2 difference, implying normal lung function. These patients, who had been included in the study in order to establish the PaCO2PE'CO2 in the absence of congenital heart disease, had diagnoses of aortic stenosis, pulmonary stenosis, patent ductus arteriosus and coarctation of the aorta, which were not expected to affect gas exchange. There were no patients in our study group with similarly uncomplicated lesions.

To find a further explanation for the much increased alveolar deadspace in some apparently well-oxygenated children with congenital heart disease, oxygen saturation while breathing spontaneously in air was plotted against oxygen saturation obtained during controlled ventilation with an increased inspired oxygen concentration (Fig. 4). Most of the children showed an improvement in saturation, implying a degree of ventilation–perfusion mismatch, other than a pure right-to-left shunt effect. In the well-oxygenated patients with significant left-to-right shunting, e.g. those with large ventricular septal defects, pulmonary congestion may have been responsible for the increased alveolar deadspace. Patients whose clinical history was suggestive of moderate to severe pulmonary congestion, typically those with a significant left-to-right shunt, are plotted separately in Fig. 3 to show their influence on the results of the group. Increased pulmonary blood flow causes loss of homeostasis at the alveolar/capillary level and a ventilation/perfusion mismatch.7 Increased slopes of phase III of the Single Breath Test for carbon dioxide, a sign of increased spread of ventilation/perfusion ratios, can be seen even in children with such relatively mild lesions as atrial septal defect.7 Very steep phase III slopes have been observed in children with congestive failure.7



View larger version (17K):
[in this window]
[in a new window]
 
Fig 4 Oxygen saturation when spontaneously breathing air plotted against arterial saturation with increased FIO2 and controlled ventilation under general anaesthesia, for 50 children with congenital heart disease.

 
In addition to these factors, we also explored the possibility that our carbon dioxide analyser had underestimated expired carbon dioxide concentration, in spite of our attempts to sample within the proximal endotracheal tube in small subjects. To investigate this hypothesis, we ventilated a carbon dioxide-producing lung model at different tidal volumes using the ventilator used in the study. This model showed a small carbon dioxide analyser error at tidal volumes below 50 ml, as shown in Fig. 5. This error was not sufficient to explain the observed PaCO2PE'CO2 in our patients with normal saturations.



View larger version (10K):
[in this window]
[in a new window]
 
Fig 5 The error in measured end-tidal PCO2 associated with small tidal volumes, using a carbon dioxide-producing lung model and a calibration test gas. The y axis represents the actual value recorded divided by the PCO2 measured at a control tidal volume of 300 ml.

 
In conclusion, the theoretical model for prediction of PaCO2PE'CO2 due to right-to-left shunting in congenital heart disease underestimated the true extent of the derangement of gas exchange in many patients. In addition to the primary, obligatory effect of the shunt, carbon dioxide elimination may be affected by pulmonary hypoperfusion in right-to-left shunts, or by congested pulmonary vasculature in mixed and left-to-right shunts. Hence, it appears that PaCO2PE'CO2 in children with congenital heart disease is difficult to predict, especially in the presence of congestive cardiac failure.

In the light of our findings, we suggest that the relationship between pulmonary congestion and PaCO2PE'CO2 could be further investigated by performing a similar study on acyanotic children with congenital heart disease, who have been prospectively classified into normal and plethoric groups, by blinded reporting of chest radiographs, elements of the clinical history and the saturation response to increased FIO2.


    References
 Top
 Abstract
 Introduction
 Theory
 Methods
 Results
 Discussion
 References
 
1 Badgwell JM, Mcleod ME, Lerman J, Creighton RE. End-tidal PCO2 measurements sampled at the distal and proximal ends of the endotracheal tube in infants and children. Anesth Analg 1987; 66: 959–64[Abstract]

2 Rich GF, Sconzo JM. Continuous end-tidal CO2 sampling within the proximal endotracheal tube estimates arterial CO2 tension in infants. Can J Anaesth 1991; 38: 201–3[Abstract]

3 Fletcher R. The relationship between the arterial to end-tidal PCO2 difference and haemoglobin saturation in patients with congenital heart disease. Anesthesiology 1991; 75: 210–6[ISI][Medline]

4 Fletcher R. Gas exchange during anaesthesia and controlled ventilation in children with congenital heart disease. Paediatr Anaesth 1993; 3: 5–17

5 Nunn JF. Nunn’s Applied Respiratory Physiology, edn 4. Oxford: Butterworth-Heinemann, 1993; 225

6 Nunn JF. Nunn’s Applied Respiratory Physiology, edn 4. Oxford: Butterworth-Heinemann, 1993; 224

7 Fletcher R, Niklason L, Drefeldt B. Gas exchange during controlled ventilation in children with normal and abnormal pulmonary circulation: a study using the single breath test for carbon dioxide. Anesth Analg 1986; 65: 645–52[Abstract]