1491
Effects of indomethacin on cerebral haemodynamics in very preterm infants
infrared spectroscopy was used to the effects of intravenously investigate administered indomethacin (0·1-0·2 mg/kg) on cerebral haemodynamics and oxygen delivery in 13 very preterm infants treated for patent ductus arteriosus. 7 infants received indomethacin by rapid injection (30 s) and 6 by slow infusion (20-30 min). In all the infants cerebral blood flow, oxygen delivery, blood volume, and the reactivity of blood volume to changes in arterial carbon dioxide tension fell sharply after indomethacin. There were no differences in the effects of rapid and slow infusion. These falls in cerebral oxygen delivery and the disruption of cerebrovascular control might compromise cellular oxygen availability, particularly in regions of the brain where the arterial supply is precarious. Care should be taken to ensure that oxygen delivery is optimum before the administration of indomethacin to preterm infants.
Patients and methods
Near
Introduction Intravenous indomethacin is commonly used for the treatment of patent ductus arteriosus in small sick infants1 and a role for the drug in the prevention of periventricular haemorrhage by stabilising the cerebral circulation has been suggested. It is important to understand the cerebrovascular effects of this drug. There are many data from experiments on animals, but differences among species make extrapolation to the human infant difficult. In the human adult, bolus injection of indomethacin causes reductions in cerebral blood flow3 and in its response to changes in arterial carbon dioxide tension (PaCOz).4 Most studies of newborn infants have used doppler ultrasound methods to measure cerebral artery blood velocity. 5-7 The results are difficult to interpret, especially since indomethacin may change the calibre of the insonated vessel, but it has been inferred that cerebral blood flow (as well as blood velocity) is reduced. A small study of cerebral blood flow measured by the xenon-133 method8 showed a fall in blood flow after rapid injection of indomethacin (02 mg/kg). There is concern that cerebral blood flow may fall to unacceptably low levels5-8 and it has been suggested that the reduction may be abolished if the drug is given by slow infusion rather than by rapid injection.9 We have used near infrared spectroscopy (NIRS) to investigate the effects of indomethacin on cerebral blood flow, oxygen delivery, blood volume, and the response of cerebral blood volume to changes in PACO, in a group of very preterm infants who required treatment for patent ductus arteriosus. The effects of both rapid and slow administration of the drug were tested.
We studied 13 very preterm infants (8 male, 5 female) admitted to the neonatal unit of our hospital (table I). On cranial ultrasound 6 infants (4 rapid, 2 slow group) had normal scans; 5 (2 rapid, 3 slow) had mild ventricular dilatation;1infant in the rapid infusion group had mild ventricular dilatation and germinal layer haemorrhage; and 1 infant in the slow infusion group had a unilateral intraventricular and small parenchymal haemorrhage. The diagnosis of clinically significant patent ductus arteriosus was made when a characteristic cardiac murmur was associated with bounding peripheral pulses and an enlarged liver; all infants had echocardiographic evidence of an aortic root/left atrial ratio of less than 0-67. The attending clinical staff decided whether treatment was needed and the dose of indomethacin. Consent for the investigation was obtained from the parents before each study and the project was approved by the University College London Faculty of Clinical Sciences Committee on the Ethics of Clinical
Investigation. Details of the NIRS system have been described elsewhere." For this study we used a commercial prototype spectrophotometer (NIR 1000, Hamamatsu Photonics KK, Japan). The technique depends on absorption by the chromophores oxyhaemoglobin and deoxyhaemoglobin of near infrared light transmitted through the cranium. The Beer-Lambert law describing optical absorption in a highly scattering medium may be expressed as:
Absorption = acLB + G is the absorption coefficient of the chromophore (mmol1-1. cm- 1), c the concentration of chromophore (mmol/1), L the distance between the points where light enters and leaves the tissue (cm), B a "pathlength factor" that takes account of the scattering of light in the tissue (which causes the optical pathlength to be greater than L), and G a factor related to the geometry of the tissue.12 If L, B, and G remain constant, changes in chromophore concentration can be calculated from the change in optical absorption divided by the product of aLB. The absorption coefficients of oxyhaemoglobin and deoxyhaemoglobin are known13 and a mean value for B in preterm infants of 4-39 (SD 0’28)14 has been obtained. In this study near infrared light was carried to the infant’s head through a fibreoptic bundle, the end of which was applied to the parietal region. Transmitted light emerging from the opposite parietal region was collected by another fibreoptic bundle leading to the photomultiplier tube of the NIR 1000. The distance between the applied ends of the fibreoptic bundles (L) was measured with callipers and the optical pathlength obtained from the product of L and B. Light at six wavelengths (7971 nm, 802-5 nm, 831-2 nm, 848 nm, 866 nm, and 907 nm) was transmitted through the head, and changes in the intracerebal concentrations of oxyhaemoglobin and deoxyhaemoglobin were calculated by the
where
a
ADDRESS Departments of Paediatrics (A D. Edwards, MRCP, J S. Wyatt, MRCP, C Richardson, MSc, A. Potter, BSc, Prof E. O. R. Reynolds, FRCP), and Medical Physics and Bioengineering (M. Cope, BSc, D T Delpy, BSc), University College and Middlesex School of Medicine, London, UK. Correspondence to Dr A D. Edwards, Department of Paediatrics, University College and Middlesex School of Medicine, Rayne Institute, University Street, London WC1 E 6JJ, UK.
1492
Fig 1-Effects
of indomethacin (01 mg/kg) on cerebral circulation variables in 1 infant from each treatment group.
Infant A born at 25 wk gestation, studied aged 14 days; infant B born at 26 wk gestation, studied aged 20 days. Cerebral oxygen delivery before and after indomethacin averaged 4 55 and 2-02 ml.100 g-1.min-1 in infant A and 3-91 and 2-28 ml.1 00 g-1.min-1 in infant B
linear least squares curve-fitting technique with the absorption coefficients reported previously.13 Cerebral blood flow was measured by the Fick principle, with a small sudden change in arterial concentration of oxyhaemoglobin used as an intravascular tracer.15-17 The change was induced by giving four or five breaths of a higher inspired oxygen concentration so that SaOz rose rapidly by about 5%. Arterial oxygen tension (Pao2) measured by transcutaneous electrode was not allowed to rise above 14 kPa. Cerebral blood flow was then calculated from the changes in Sa02 and cerebral oxyhaemoglobin concentration recorded over 6-8 s, as: Cerebral blood flow
=
Change in oxyhaemoglobin concentration k —————————————————:——:———"-
(Total haemoglobin concentration in of (o-->t]
change
x
integral
SaOz)
where k=089, a constant reflecting the molecular weight of haemoglobin (64 500), tissue density (1-05), and cerebral/largevessel haematocrit ratio (0-6919). Cerebral oxygen delivery was calculated as the product of arterial oxygen content and cerebral blood flow, assuming a cerebral/largevessel haematocrit ratio of 0-69.19 Arterial oxygen content was calculated from a standard formula2O with average Sa02 measured before and after indomethacin while the infant was quiet and well
oxygenated. Cerebral blood volume was quantified by observation of the effect of a small gradual change in Sa02 on the concentrations of oxyhaemoglobin and deoxyhaemoglobin.21 This method also uses oxyhaemoglobin as a tracer molecule measured by NIRS, but the measurements were made during a much slower change in Sa02 over several minutes, so that the tracer was at steady state and equilibrated throughout the brain. The dilution of the tracer in the brain allowed the calculation of cerebral blood volume as:
Change in (oxyhaemoglobin deoxyhaemoglobin concentration) Cerebral blood volume = k
——————————————————————————
(2 x Change in SaOz x total haemoglobin concentration)
where k is as defined above. Because NIRS provides continuous information about changes in the total cerebral haemoglobin concentration the value obtained for cerebral blood volume could be used to calibrate the record to provide continuous data for that variable.
Fig 2-lndividual
mean values for cerebral circulation variables before and after indomethacin.
Cerebral blood volume response to carbon dioxide was measured
by observation of the change in cerebral blood volume in response to a spontaneous or induced (by changes in ventilator settings) change in PACO, of about 1 kPa within or towards the normal range over several minutes. Cerebral blood volume normally changed directly with changes in PACO,. NIRS measurements were made for 80-287 (median 199) min. One to four (median one) measurements of cerebral blood flow were made 8-189 min before drug administration and between one and eight (median three) measurements 1-110 min afterwards. Changes in cerebral blood volume were measured continuously in each infant and the volume response to changes in PACO, was measured once starting 9-110 min before and once 12-42 min after injection. NIRS and other data were stored on magnetic disc for later analysis. Normal nursing and medical care continued without interference during the studies. To estimate PaOz and PaCOz continuous recordings were made of transcutaneous P02 and PCOz by means of a ’Novametrix’ 850 electrode calibrated by analysis of arterial blood. Arterial oxygen saturation was measured on every pulse beat with a modified ’Novametrix’ 500 pulse oximeter sited on the ear or the lip. Total haemoglobin concentration was measured by Coulter counter.
1493
TABLE I-CLINICAL DETAILS OF STUDY INFANTS
*1 other
was receiving continuous positive airway tlmmediately before indomethacin.
pressure.
Arterial blood pressure was measured before and after indomethacin administration by an oscillometric method (Critikon:
’Dinamap’ 1846SX). Indomethacin was dissolved in 1-2 ml 0-9% sodium chloride solution and injected over about 30 s into a peripheral vein in 7 infants (the "rapid" group). For slow administration the drug was further diluted in 0-9% sodium chloride solution to a total volume of about 5 ml and an infusion pump was used to deliver this volume over 20-30 min to 6 infants (the "slow" group). 2 infants in the rapid and 3 in the slow group received a dose of 02 mg/kg; the remainder received 0-1mg/kg. In all cases the infant was receiving the first of a course of three doses of indomethacin when studied. To provide control data we observed the effects on cerebral blood volume of seven rapid injections of the following substances, made up in volumes of between 0-1 and 2-0 ml: phenobarbitone 10 ml/kg (two observations); 0-9% saline; morphine 0-1 mg/kg; gentamicin 2-5 mg/kg; penicillin 30 mg/kg; and pancuronium 0-005 mg (one observation each). The mean cerebral blood flow and oxygen delivery were calculated before and after indomethacin for each infant. Cerebral blood volume before indomethacin was defined as the average volume in the 5 min immediately before drug administration; that afterwards was the average volume in the 5 min starting 5 min after the injection in the rapid group and 30 min after the infusion started in the slow group. To ascertain that these values were representative of the changes induced by indomethacin, they were compared with the maximum change recorded and no significant differences were found. A paired t test was used to test individual mean values for cerebral blood flow, oxygen delivery, blood volume, and response to changes in PaCO for significant differences before and after indomethacin. Changes in these variables were compared between groups by the unpaired t test. Non-parametric tests were also used, but the results were similar. The power of the study to detect differences between rapid and slow groups was estimated.
TABLE II-GROUP MEAN (SD) VALUES OF CEREBRAL CIRCULATION VARIABLES BEFORE AND AFTER INDOMETHACIN
All values after indomethacin those before.
were
significantly (p
Effects of indomethacin on cerebral haemodynamics in very preterm infants
infrared spectroscopy was used to the effects of intravenously investigate administered indomethacin (0·1-0·2 mg/kg) on cerebral haemodynamics and oxygen delivery in 13 very preterm infants treated for patent ductus arteriosus. 7 infants received indomethacin by rapid injection (30 s) and 6 by slow infusion (20-30 min). In all the infants cerebral blood flow, oxygen delivery, blood volume, and the reactivity of blood volume to changes in arterial carbon dioxide tension fell sharply after indomethacin. There were no differences in the effects of rapid and slow infusion. These falls in cerebral oxygen delivery and the disruption of cerebrovascular control might compromise cellular oxygen availability, particularly in regions of the brain where the arterial supply is precarious. Care should be taken to ensure that oxygen delivery is optimum before the administration of indomethacin to preterm infants.
Patients and methods
Near
Introduction Intravenous indomethacin is commonly used for the treatment of patent ductus arteriosus in small sick infants1 and a role for the drug in the prevention of periventricular haemorrhage by stabilising the cerebral circulation has been suggested. It is important to understand the cerebrovascular effects of this drug. There are many data from experiments on animals, but differences among species make extrapolation to the human infant difficult. In the human adult, bolus injection of indomethacin causes reductions in cerebral blood flow3 and in its response to changes in arterial carbon dioxide tension (PaCOz).4 Most studies of newborn infants have used doppler ultrasound methods to measure cerebral artery blood velocity. 5-7 The results are difficult to interpret, especially since indomethacin may change the calibre of the insonated vessel, but it has been inferred that cerebral blood flow (as well as blood velocity) is reduced. A small study of cerebral blood flow measured by the xenon-133 method8 showed a fall in blood flow after rapid injection of indomethacin (02 mg/kg). There is concern that cerebral blood flow may fall to unacceptably low levels5-8 and it has been suggested that the reduction may be abolished if the drug is given by slow infusion rather than by rapid injection.9 We have used near infrared spectroscopy (NIRS) to investigate the effects of indomethacin on cerebral blood flow, oxygen delivery, blood volume, and the response of cerebral blood volume to changes in PACO, in a group of very preterm infants who required treatment for patent ductus arteriosus. The effects of both rapid and slow administration of the drug were tested.
We studied 13 very preterm infants (8 male, 5 female) admitted to the neonatal unit of our hospital (table I). On cranial ultrasound 6 infants (4 rapid, 2 slow group) had normal scans; 5 (2 rapid, 3 slow) had mild ventricular dilatation;1infant in the rapid infusion group had mild ventricular dilatation and germinal layer haemorrhage; and 1 infant in the slow infusion group had a unilateral intraventricular and small parenchymal haemorrhage. The diagnosis of clinically significant patent ductus arteriosus was made when a characteristic cardiac murmur was associated with bounding peripheral pulses and an enlarged liver; all infants had echocardiographic evidence of an aortic root/left atrial ratio of less than 0-67. The attending clinical staff decided whether treatment was needed and the dose of indomethacin. Consent for the investigation was obtained from the parents before each study and the project was approved by the University College London Faculty of Clinical Sciences Committee on the Ethics of Clinical
Investigation. Details of the NIRS system have been described elsewhere." For this study we used a commercial prototype spectrophotometer (NIR 1000, Hamamatsu Photonics KK, Japan). The technique depends on absorption by the chromophores oxyhaemoglobin and deoxyhaemoglobin of near infrared light transmitted through the cranium. The Beer-Lambert law describing optical absorption in a highly scattering medium may be expressed as:
Absorption = acLB + G is the absorption coefficient of the chromophore (mmol1-1. cm- 1), c the concentration of chromophore (mmol/1), L the distance between the points where light enters and leaves the tissue (cm), B a "pathlength factor" that takes account of the scattering of light in the tissue (which causes the optical pathlength to be greater than L), and G a factor related to the geometry of the tissue.12 If L, B, and G remain constant, changes in chromophore concentration can be calculated from the change in optical absorption divided by the product of aLB. The absorption coefficients of oxyhaemoglobin and deoxyhaemoglobin are known13 and a mean value for B in preterm infants of 4-39 (SD 0’28)14 has been obtained. In this study near infrared light was carried to the infant’s head through a fibreoptic bundle, the end of which was applied to the parietal region. Transmitted light emerging from the opposite parietal region was collected by another fibreoptic bundle leading to the photomultiplier tube of the NIR 1000. The distance between the applied ends of the fibreoptic bundles (L) was measured with callipers and the optical pathlength obtained from the product of L and B. Light at six wavelengths (7971 nm, 802-5 nm, 831-2 nm, 848 nm, 866 nm, and 907 nm) was transmitted through the head, and changes in the intracerebal concentrations of oxyhaemoglobin and deoxyhaemoglobin were calculated by the
where
a
ADDRESS Departments of Paediatrics (A D. Edwards, MRCP, J S. Wyatt, MRCP, C Richardson, MSc, A. Potter, BSc, Prof E. O. R. Reynolds, FRCP), and Medical Physics and Bioengineering (M. Cope, BSc, D T Delpy, BSc), University College and Middlesex School of Medicine, London, UK. Correspondence to Dr A D. Edwards, Department of Paediatrics, University College and Middlesex School of Medicine, Rayne Institute, University Street, London WC1 E 6JJ, UK.
1492
Fig 1-Effects
of indomethacin (01 mg/kg) on cerebral circulation variables in 1 infant from each treatment group.
Infant A born at 25 wk gestation, studied aged 14 days; infant B born at 26 wk gestation, studied aged 20 days. Cerebral oxygen delivery before and after indomethacin averaged 4 55 and 2-02 ml.100 g-1.min-1 in infant A and 3-91 and 2-28 ml.1 00 g-1.min-1 in infant B
linear least squares curve-fitting technique with the absorption coefficients reported previously.13 Cerebral blood flow was measured by the Fick principle, with a small sudden change in arterial concentration of oxyhaemoglobin used as an intravascular tracer.15-17 The change was induced by giving four or five breaths of a higher inspired oxygen concentration so that SaOz rose rapidly by about 5%. Arterial oxygen tension (Pao2) measured by transcutaneous electrode was not allowed to rise above 14 kPa. Cerebral blood flow was then calculated from the changes in Sa02 and cerebral oxyhaemoglobin concentration recorded over 6-8 s, as: Cerebral blood flow
=
Change in oxyhaemoglobin concentration k —————————————————:——:———"-
(Total haemoglobin concentration in of (o-->t]
change
x
integral
SaOz)
where k=089, a constant reflecting the molecular weight of haemoglobin (64 500), tissue density (1-05), and cerebral/largevessel haematocrit ratio (0-6919). Cerebral oxygen delivery was calculated as the product of arterial oxygen content and cerebral blood flow, assuming a cerebral/largevessel haematocrit ratio of 0-69.19 Arterial oxygen content was calculated from a standard formula2O with average Sa02 measured before and after indomethacin while the infant was quiet and well
oxygenated. Cerebral blood volume was quantified by observation of the effect of a small gradual change in Sa02 on the concentrations of oxyhaemoglobin and deoxyhaemoglobin.21 This method also uses oxyhaemoglobin as a tracer molecule measured by NIRS, but the measurements were made during a much slower change in Sa02 over several minutes, so that the tracer was at steady state and equilibrated throughout the brain. The dilution of the tracer in the brain allowed the calculation of cerebral blood volume as:
Change in (oxyhaemoglobin deoxyhaemoglobin concentration) Cerebral blood volume = k
——————————————————————————
(2 x Change in SaOz x total haemoglobin concentration)
where k is as defined above. Because NIRS provides continuous information about changes in the total cerebral haemoglobin concentration the value obtained for cerebral blood volume could be used to calibrate the record to provide continuous data for that variable.
Fig 2-lndividual
mean values for cerebral circulation variables before and after indomethacin.
Cerebral blood volume response to carbon dioxide was measured
by observation of the change in cerebral blood volume in response to a spontaneous or induced (by changes in ventilator settings) change in PACO, of about 1 kPa within or towards the normal range over several minutes. Cerebral blood volume normally changed directly with changes in PACO,. NIRS measurements were made for 80-287 (median 199) min. One to four (median one) measurements of cerebral blood flow were made 8-189 min before drug administration and between one and eight (median three) measurements 1-110 min afterwards. Changes in cerebral blood volume were measured continuously in each infant and the volume response to changes in PACO, was measured once starting 9-110 min before and once 12-42 min after injection. NIRS and other data were stored on magnetic disc for later analysis. Normal nursing and medical care continued without interference during the studies. To estimate PaOz and PaCOz continuous recordings were made of transcutaneous P02 and PCOz by means of a ’Novametrix’ 850 electrode calibrated by analysis of arterial blood. Arterial oxygen saturation was measured on every pulse beat with a modified ’Novametrix’ 500 pulse oximeter sited on the ear or the lip. Total haemoglobin concentration was measured by Coulter counter.
1493
TABLE I-CLINICAL DETAILS OF STUDY INFANTS
*1 other
was receiving continuous positive airway tlmmediately before indomethacin.
pressure.
Arterial blood pressure was measured before and after indomethacin administration by an oscillometric method (Critikon:
’Dinamap’ 1846SX). Indomethacin was dissolved in 1-2 ml 0-9% sodium chloride solution and injected over about 30 s into a peripheral vein in 7 infants (the "rapid" group). For slow administration the drug was further diluted in 0-9% sodium chloride solution to a total volume of about 5 ml and an infusion pump was used to deliver this volume over 20-30 min to 6 infants (the "slow" group). 2 infants in the rapid and 3 in the slow group received a dose of 02 mg/kg; the remainder received 0-1mg/kg. In all cases the infant was receiving the first of a course of three doses of indomethacin when studied. To provide control data we observed the effects on cerebral blood volume of seven rapid injections of the following substances, made up in volumes of between 0-1 and 2-0 ml: phenobarbitone 10 ml/kg (two observations); 0-9% saline; morphine 0-1 mg/kg; gentamicin 2-5 mg/kg; penicillin 30 mg/kg; and pancuronium 0-005 mg (one observation each). The mean cerebral blood flow and oxygen delivery were calculated before and after indomethacin for each infant. Cerebral blood volume before indomethacin was defined as the average volume in the 5 min immediately before drug administration; that afterwards was the average volume in the 5 min starting 5 min after the injection in the rapid group and 30 min after the infusion started in the slow group. To ascertain that these values were representative of the changes induced by indomethacin, they were compared with the maximum change recorded and no significant differences were found. A paired t test was used to test individual mean values for cerebral blood flow, oxygen delivery, blood volume, and response to changes in PaCO for significant differences before and after indomethacin. Changes in these variables were compared between groups by the unpaired t test. Non-parametric tests were also used, but the results were similar. The power of the study to detect differences between rapid and slow groups was estimated.
TABLE II-GROUP MEAN (SD) VALUES OF CEREBRAL CIRCULATION VARIABLES BEFORE AND AFTER INDOMETHACIN
All values after indomethacin those before.
were
significantly (p
