Acta Padiatr 81: 498-503. 1992

Cardiovascular effects of carbon dioxide in ventilated preterm infants AC Fenton, K L Woods’, R Leanage, M Abu-Harb, MI Levene, D H Evans2 and DJ Field Departments of Child Health. Pharmacology‘ and Medical Physics2, Leirester University Medical School, Leicester. U K

Fenton AC, Woods KL, Leanage R, Abu-Harb M, Levene MI, Evans DH, Field DJ. Cardiovascular effects of carbon dioxide in ventilated preterm infants. Acta Pzdiatr 1992;81:498-503. Stockholm. ISSN 0803-5253 Sick preterm infants may, under certain conditions, demonstrateblood pressure passive cerebral blood flow in response to changes in arterial carbon dioxide tension. Blood pressure in turn depends on cardiac output and peripheral resistance. A Doppler technique for assessing cardiac output compared favourably in terms of reproducibilily to a thermodilutiontechnique in a group of infants undergoing cardiac catheterization for congenital heart disease. Doppler was subsequently used to monitor changes in cardiac output following an increase in arterial carbon dioxide tension of 1 kPa in 25 ventilated preterm infants. Blood pressure increased significantly ( p =0.006). However, heart rate did not change significantly ( p = 0.16) and, in addition, both stroke and minute volume decreased ( p = 0.023, p = 0.02, respectively). This suggests that accompanying changes in components of peripheral resistance exert important effects on blood pressure in the preterm neonate in response to changes in arterial carbon dioxide tension. 0 Carbon dioxide, Doppler ultrasound, preterm infants DH Evans, Department of Medical Physics, Leicester Royal Infirmary. Infirmary Square, Leics LEI 5 W W , U K

Recent work ( I ) has suggested that under certain conditions cerebral blood flow in sick preterm infants is directly influenced by arterial blood pressure. This has been hypothesized to predispose these infants to neurodevelopmental injury (2). Clearly, an understanding of the factors which influence blood pressure in these infants is important. Blood pressure is the product of peripheral resistance and cardiac output. There is no method available for quantitative non-invasive assessment of components of overall vascular resistance, such as the renal and mesenteric beds, and any such measurements in the neonate would be further complicated by the small size of vessels involved. Measurement of cardiac output, an important expression of cardiac function, also presents difficulties both in method and in the interpretation of the results obtained if it is measured relatively infrequently. Currently, thermodilution is taken as being a “goldstandard” method, but there are several problems with this technique, both in general terms and in its use in paediatric populations (3). Indeed the validity of using this method to determine the validity of other techniques has been questioned (4). Several previous studies have compared thermodilution with Doppler determination of cardiac output in a paediatric population (5-7), and achieved varying degrees of correlation. Most recently, in a study of 17 children receiving intensive care, Notterman et al. (8) concluded that non-imaging Doppler assessment of cardiac output was not sufficiently accurate to be used

instead of thermodilution. The authors wondered whether the use of concurrent imaging of the aorta would improve accuracy. We wanted to use a Doppler method to monitor changes in cardiac output in preterm neonates following a rise in arterial carbon dioxide tension (PaC02) of approximately 1 kPa as a progression of a previous study (1). The reproducibility of Doppler measurements (9) lends itself to observing such intra-individual changes. We wished to use a duplex system, which allows reproducible placement of the Doppler sample volume, in addition to minimizing the angle of insonation of the vessel which is so critical to Doppler shift calculations. Owing to the variability of correlation between thermodilution and Doppler in the literature, we felt it important to perform our own validation study prior to using the technique in neonates to determine its reproducibility in our own hands.

Validation of cardiac output measurements Patients

The validation study group comprised 10 infants (age range 1-11 months, mean 4.4 months) undergoing cardiac catheterization as part of their management for congenital cardiac disease. All patients had either normal cardiac anatomy (i.e. postoperative) or a septa1 defect. Full clinical details are given in Table 1. All catheter studies were performed using intramuscular pethidine compound (0.1 ml/kg im) for sedation. A 5 F thermodilution catheter (Spectramed SP5 105H)

1.33

1.54

2.27

1.13

1.33

1.13

1.33

11

3

3

1

4

2

6

7

Arterial switch for simple TGA

VSD ? Coarctation

VSD

Arterial switch for simple TGA

Repaired coarctation (normal intra-cardiac anatomy)

Arterial switch for simple TGA

Patient foramen ovale

VSD

3

4

5

6

7

8

9

10

1.33

1.13

3

2

ASD

VSD

I

1.13

Diagnosis

Aortic CSA (cm2)

4

Infant

Age (months) -

19.4 20.5 16.8

15.6 17.0 9.2 10.5 8.0 16.2 15.7 15.6

15.4

19.8 18 2 16.8 16.3 16.2 17.3 18.8 18.7 17.7 13.3 10.0 10.5 4.1 9.4 6.1 7.0 7.8 7.6 8.0 .90

i:g} I

2.23

X?}

1.06 1.83 1.77} 1.76

1.92 1.22 1.40}

M7}

1.06

1.59

"13] I .38

1.62 0.93

I: 2.35

;:it}

2.51

1.79

2.41} 2.17 2.12

:is}

::%]

0.93

2.20

1.19

}!;:

1.80

1.71) 1.81

in}

ki}

1.81 1.521 1.63 1.82 1.23

1.51

1.63

2.45

1.87

2.06

D

2.65

1.83

1.26

1.66

0.90

1.92

1.65

1.77

1.36

1.11

T

Thermodilution

Cardiac output (I/min) Doppler

} : : I :1.95

Minute distance (m)

2.58

1.81

1.25

1.74

0.97

1.72

1.64

2.11

1.62

1.59

Mean D, T

-0.14

-0.04

-0.03

0.15

0.14

-0.41

-0.02

0.68

0.51

0.95

Difference D-T

2.66

1.80

1.31

1.75

1.03

I .53

I .64

2.50

1.84

2.15

Mean

-0.15

0.06

-0.18

-0.02

0.03

-1.20

0.20

0.01

0.01

0.19

2.59

1.86

1.27

1.57

0.89

1.79

1.63

1.76

1.36

1.19

Difference Mean

-0.36

0.19

0.04

-0.11

-0.02

-1.20

0.19

-0.10

0.08

0.04

Difference

Paired estimations Paired estimations (Doppler) (thermodilution)

Table I. Doppler/thermodilution validation. D&T refer to mean of replicate readings in each infant for Doppler and thermodilution, respectively. Paired estimations refer to the first pair of values in each set of replicate measurements.

500

ACTA PRDIATR 81 (1992)

AC Fenron et al.

was inserted percutaneously under local anaesthesia into one femoral vein and positioned either in the pulmonary artery (if cardiac anatomy was normal) or manipulated into the aortic root (if a septal defect was present). The position of the catheter was confirmed radiographically. Three thermodilution cardiac outputs using 5 ml of chilled dextrose 5% solution were performed in each patient at 1-min intervals, using a Gould SP1435 Cardiac Index Computer. The system incorporates an in-line temperature probe to ensure suitable injectate temperatures. All catheter studies were performed by the same operator (RL). Immediately following each thermodilution measurement, a Doppler estimation of stroke distance was performed by one of the authors (AF). The ascending aorta was visualized from the suprasternal notch, using an ATL 600 duplex Doppler system with a 724A probe (pulse-echo frequency 7.5 MHz, Doppler frequency 5.0 MHz). The Doppler sample volume was positioned in the centre of the aortic root, care being taken to ensure that the angle of insonation of the vessel was less than 10”. For all recordings the Doppler sample volume was set at 5 mm and the high-pass wall filter at its lowest value (100 Hz). The ultrasound intensity was kept as low as possible and never allowed to exceed 100 mW/cm2 SPTA. Once the optimum Doppler signal had been obtained by listening to the audio signal and observing the sonogram in real time, 25 to 30 cardiac cycles were recorded on to digital audiotape for analysis. The probe was removed between each determination. Doppler analysis The Doppler tapes were replayed through a fast Fourier transform analyser (10) and the maximum frequency envelope of 20 or more consecutive beats was extracted and calibrated in terms of velocity using the Doppler equation. Stroke distance was calculated from the timeaveraged maximum velocity, by assuming flow in the aorta to be “plug-like’’ (1 1). The error introduced by the fact that the flow profile is probably somewhere between plug-like and parabolic, thus altering the relationship between maximum and mean velocity, would not affect assessment of reproducibility of the technique. Minute distance was calculated as the product of stroke distance and heart rate. Aortic diameter measurements for crosssectional area estimations were made from real-time ultrasound images. Cardiac output was calculated as the product of minute distance and aortic cross-sectional area, the latter being assumed to be circular. This assumption would also not affect reproducibility.

0.50

0.25

0.00 -0.25 -0.50

-0.75

0.0

0.5

1.0

1.5

2.0

3.0

2.5

Mean Estimate of Cardiac Output (I.rnln

’)

Fig. 1. Correlation of Doppler and thermodilution (plotted as difference between the methods against their mean).

Doppler and thermodilution estimations. The mean of these two values for each infant was plotted against the difference between them (Table 1 & Fig. 1). The mean difference in output estimated by the two methods was 0.18 l/min (SD 0.39 l/min), which clearly shows a wide range with poor agreement. The aim of the study, however, was to assess the reproducibility of the two techniques, and this question was addressed using the method of Bland & Altman (12). The first pair of estimates of cardiac output were taken for each method in each infant. The difference between the two estimations was plotted against their mean (Table 1 & Fig. 2). One infant with a ventriculoseptal defect demonstrated very poor reproducibility for both methods (possibly related to activity and changes

9

A

Doppler Thermodilutlon

t

0

Statistical analysis Initially a direct comparison between cardiac output obtained from the two methods was attempted. A mean output in each infant was calculated for both the

1

1

2

Mean Estimate of Cardiac Output (I.min

’

3 )

Fig. 2. Reproducibility of both Doppler (0) and thermodilution ( A ) . One infant (filled symbols) demonstrated poor reproducibility for both methods.

Preterm cardiovascular response to COz

ACTA PEDIATR XI (1992)

in the degree of intra-cardiac shunting during the procedure) and was therefore excluded from subsequent analysis. The remaining infants showed no significant differences in reproducibility, the S D of the differences being 0.129 l/min for Doppler and 0.171 l/min for thermodilution. Thus, despite the poor correlation between the two methods, the reproducibility of both was good, the Doppler method performing a t least as well as the thermodilution method. It was therefore felt appropriate to apply the technique to estimating changes in stroke distance in ventilated preterm infants.

Effect of PaC02 rise on cardiac output in preterm infants Patients and methods The study was performed in the neonatal unit, Leicester Royal Infirmary. All infants of 134 weeks gestation who were ventilated for idiopathic respiratory distress syndrome were considered eligible. It is our practice for all such infants to have indwelling arterial catheters (either umbilical or peripheral) for blood-gas sampling and blood pressure monitoring. Infants were not studied if their condition was considered unstable, if their initial PaC02 was 2 7 kPa, or if there was no indwelling arterial line. Infants who had previously developed a pneumothorax during the course of their illness were excluded, in view of potential compromise of cardiac function. The study was approved by the Leicestershire District Ethics Committee and informed parental consent was obtained prior to enrolling infants in the study. Study procedure Prior to each study the following parameters were recorded: Blood pressure. Blood pressure was measured by a pressure transducer (Spectramed P23XL) attached via non-compliant manometer tubing to the infant’s indwelling arterial line as described previously ( 1 3). A permanent record of blood pressure was made on a chart recorder (Gould) and mean arterial blood pressure (MABP) noted. Stroke distance. As in the validation study, the ascending aorta was visualized from the suprasternal notch, using an identical Doppler system. Two determinations of stroke distance 10 min apart were made as previously, the probe being removed between each recording. Analysis of the Doppler tapes was also performed in the same way. Ventilatory parameters. Ventilator settings and whether the infant was paralysed with pancuronium at the time of the study were recorded. Prior to each study, 0.2 ml of blood was taken from the infant’s arterial line for PaCOz and arterial oxygen tension (PaO2) determinations. Routine sedation with an opiate was not used during the course of this study.

50 1

COZchallenge

The PaC02 was increased by placing a 5-15 ml deadspace (depending on the weight of the infant) into the ventilator circuit between the ventilator manifold and the end of the endotracheal tube. The deadspace was made of wide-bore non-compliant tubing and was shown to be of negligible resistance compared to the endotracheal tube. To confirm that the use of such a deadspace did not affect the ventilator pressures being delivered, airway pressure was measured using a butterfly needle inserted into the endotracheal tube, attached to a pressure transducer (Spectramed P23XL). The PaCOz was then allowed to equilibrate for 10-15 min, the change being monitored where possible by a transcutaneous CO2 electrode. A further 0.2 ml of blood was taken for repeat blood-gas analysis, and then repeat MABP and stroke distance recordings were made as before. Statistical analysis As a normal distribution of the data could not be assumed, Wilcoxon’s signed rank test was used for analysis of differences in blood pressure, heart rate, stroke and minute distance before and after an increase in PaC02. The effect of these differences on gestational and postnatal age and whether the infant was paralysed with pancuronium was also examined using the MannWhitney test for unpaired data. For the purposes of the latter analyses, the effect of gestational age was examined by dividing the infants into two groups; those I 30 weeks gestation (n= 1 I ) and those 2 3 1 weeks. The effect of postnatal age was examined by dividing the infants into those studied I 2 4 h of age (n =9) and those studied 2 2 5 h.

Results Twenty-one infants (gestational age 24-34 weeks, mean 30 weeks) were studied on a total of 45 occasions, 12 infants being studied at least twice. Postnatal age at the time of study ranged from 5 to 1 15 h (mean 38 h). There was no significant difference between the two baseline values for heart rate ( p = 0.59), stroke distance ( p = 0.44), minute distance ( p = 0.20) or blood pressure ( p = 0.85). Mean baseline values were as follows: PaC02 4.70 kPa (range 3.52-6.68 kPa), MABP 41 mmHg (range 30-57 mmHg), heart rate 147 beats/min (range 121-176 beats/min). The mean increase in PaC02 induced was 1.02 kPa (SEM 0.07 kPa), which resulted in a significant increase in MABP (p=0.006). The mean increase in MABP corrected for a 1-kPa rise in PaC02 was 3.7 mmHg (SEM 1.1 mmHg), which represented a mean rise of 8% from baseline values. This increase was unaffected by gestation (p=0.5), the use of paralysing agents (p=O.28) or postnatal age (p=O.46). However, for the

502 AC Fenton ei al.

ACTA PRDIATR 81 (1992)

t I

I

Pre

post C 0 2 Challenge

Fig. 3. Decrease in minute distance (m) for each infant on the first occasion studied.

group as a whole the change in heart rate was not significant (p=O.16) and a decrease in both stroke and minute distance occurred ( p = 0.023 and p = 0.02, respectively). The changes in minute distance are shown in Fig. 3, and it can be seen that the response was heterogeneous, with six infants demonstrating an increase in minute distance. Four of these six infants were therapeutically paralysed with pancuronium and had more severe respiratory disease at the time of study, but clearly this subgroup is too small to draw further conclusions. For the group as a whole, the mean decrease in minute distance for a 1-kPa increase in PaC02 (1.59 (SEM 0.63) m), represented a 10% change from baseline values and was also unaffected by gestation ( p = 0.70), the use of paralysing agents (p=O.21) or postnatal age (p=O.41). An increase in Pa02 (mean 0.26 (SEM 0.53) kPa) in these infants occurred as in our previous study (1) but did not attain statistical significance for the group as a whole ( p = 0.64).

Discussion The findings of this study that in response to the increase in PaC02,blood pressure increased despite no change in heart rate combined with a decrease in stroke volume were unexpected. Archer et al. (14) also reported no change in heart rate following an increase in PaC02 in term infants, but blood pressure was not recorded in

that study. Whilst C02 is known to have a depressant action on isolated or beta-blocked heart preparations, reducing both rate and force of contraction, in the presence of intact autonomic innervation, the sympathetic response to the rise in PaC02 results in a net increase in cardiac output in both human adult and animal studies (1 5, 16). These findings, however, suggest that the increase in blood pressure in this situation depends on an increase in peripheral resistance. Cullen & Eger (17) demonstrated a 7% increase in blood pressure per I-kPa increase in PaC02 in healthy adults, but in contrast, heart rate and stroke volume increased by 21 % and 8%, respectively, in their subjects and peripheral resistance (measured in the limb) decreased by 13%. The vasodilatory effects of increases in PaC02 on limb and splanchnic vascular beds are well documented in animal studies (15); there is a more marked response in the absence of autonomic innervation or with the use of sympathetic blockade (18). The reasons for the difference between the preterm and the adult response are not clear. It may reflect incomplete maturation of autonomic innervation of the heart in the former, thus giving more importance to factors such as vascular innervation and circulating catecholamines. The basal blood flow to the brain in the neonate is proportionately greater than in the adult and, as in the adult, a decrease in cerebrovascular resistance accompanies an increase in PaC02 (19). This means that for total peripheral resistance to increase there must be a considerable increase in peripheral resistance in other areas of the circulation. In summary, we have demonstrated that Doppler provides a simple non-invasive means of assessing changes in cardiac output in infants. The neonatal data suggest that ventilated preterm infants show a different cardiovascular response to changes in PaC02 to that in the adult. Clearly this is an area which requires further evaluation. Acknowledgement.-ACF is supported by the Spastics Society.

References 1. Fenton AC, Field DJ, Woods KL, Evans DH, Levene MI.

2. 3.

4. 5.

6.

Circulatory effects of fast ventilator rates in preterm infants. Arch Dis Child 1990;65:662-6 Wigglesworth JS, Pape KE. An integrated model for haemorrhagic and ischaemic lesions in the newborn brain. Early Hum Dev 1978;2/2:179-99 MacKenzie JD, Haites NE, Rawles JM. Method of assessing the reproducibility of blood flow measurement: factors influencing the performance of thermodilutioncardiac output computers. Br Heart J 1986;55:14-24 Bernstein DP. Noninvasive cardiac output, Doppler flowmetry, and gold-plated assumptions. Crit Care Med 1987;15:886-8 Alverson DC, Eldridge M, Dillon T, Yabek SM, Berman W Jr. Noninvasive pulsed Doppler determination of cardiac output in neonates and children. J Pediatr 1982:101:46-50 Walther FJ, Siassi B, Ramadan NA, Ananda AK, Wu PYK. Pulsed Doppler determinations of cardiac output in neonates: normal standards for clinical use. Pediatrics 1985;76829-33

Prererm cardiovascular response

ACTA PEDIATR 81 (1992)

7. Scholler GF, Whight CM, Celermajer JM. Pulsed Doppler echocardiographic assessment, including use of aortic leaflet separation, of cardiac output in children with structural heart disease. Am J Cardiol 1986;57:1195-7 8. Notterman DA, Castello FV, Steinberg C, Greenwald BM, OLoughlin JE, Gold JP. A comparison of thermodilution and pulsed Doppler cardiac output in critically ill children. J Pediatr 1989;115:554-60 9. Chandraratna PA, Nanna M, McKay C, et al. Determination of cardiac output by transcutaneous continuous wave ultrasonic Doppler computer. Am J Cardiol 1984;53:234-7 10. Schlindwein FS, Smith MJ, Evans DH. Spectral analysis of Doppler signals and computation of the normalised first moment in real time using a digital signal processor. Med Biol Eng Comput 1988;26:228-32 11. Seed WA, Wood NB. Velocity patterns in the aorta. Cardiovasc Res 1971 5 319-30 12. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;i:307- 10 13. Evans DH, Lark GM, Archer LNJ, Levene MI. The continuous measurement of intra-arterial pressure in the neonate: method and accuracy. Clin Phys Physiol Meas 1986;7:179-84

10 CO2

503

14. Archer LNJ, Evans DH, Paton JY, Levene MI. Controlled hypercapnia and neonatal cerebral Doppler waveforms. Pediatr Res 1986;20218-21 15. Price HL. Effects of carbon dioxide on the cardiovascular system. Anesthesiology 1960;21:652-63 16. Norman J, Atkinson SA. The effect of cardiac sympathetic blockade on the relationship between cardiac output and carbon dioxide tension in anaesthetized dogs. Br J Anaesth 1970;42:592602 17. Cullen DJ. Eger EI. Cardiovascular effects of carbon dioxide in man. Anesthesiology 1974;41:345-9 18. Richardson DW, Wasserman AJ, Patterson JL Jr. General and regional circulatory responses to change in blood pH and carbon dioxide tension. J Clin Invest 1961;403143 19. Levene MI, Shortland D, Gibson N, Evans DH. Carbon dioxide reactivity of the cerebral circulation in extremely premature infants: effects of postnatal age and indomethacin. Pediatr Res 1988;24:175-9

.

Received Dec. 13, 1990. Accepted Aug 29, 1991