Acta Pædiatrica ISSN 0803-5253
REGULAR ARTICLE
Effect of carbon dioxide on cerebral blood flow velocity in preterm infants during postnatal transition Shahab Noori ([email protected])1, Michael Anderson2, Sadaf Soleymani1,3, Istvan Seri1 1.Division of Neonatology and the Center for Fetal and Neonatal Medicine, Department of Pediatrics, Children’s Hospital Los Angeles and the LAC+USC Medical Center, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA 2.Biostatistics and Epidemiology, University of Oklahoma Health Sciences Center, Oklahoma City, OK, USA 3.Department of Biomedical Engineering, Viterbi School of Engineering, University of Southern California, Los Angeles, CA, USA
Keywords Haemodynamics, Middle cerebral artery, Neonate, Ultrasound, Vascular Doppler Correspondence S Noori, MD, Children’s Hospital Los Angeles, 4650 Sunset Blvd, MS# 31, Los Angeles, CA 90027, USA. Tel: (323) 361 5939 | Fax: (323) 361 7927 | Email: [email protected] Received 25 November 2013; revised 18 February 2014; accepted 25 March 2014. DOI:10.1111/apa.12646
ABSTRACT Aim: High arterial carbon dioxide (PaCO2) and cerebral reperfusion are associated with peri/intraventricular haemorrhage. Our aim was to study the relationship between PaCO2 and cerebral blood flow (CBF) in preterm infants during postnatal transition. Methods: We prospectively studied ≤30 weeks’ gestation haemodynamically stable preterm infants during the first three postnatal days (n = 21; gestational age 25.8 1.4 weeks). We measured middle cerebral artery mean flow velocity (MCA-MV) as a surrogate for CBF at the time of blood gas analysis. Results: We obtained 78 PaCO2–MCA-MV data pairs. The expected positive linear relationship between PaCO2 and MCA-MV was absent on the first postnatal day, equivocal on the second and present on the third. Using piecewise bilinear regression models, we identified PaCO2 breakpoints at 52.7 and 51.0 mmHg for postnatal days two and three, respectively. Conclusion: In haemodynamically stable preterm neonates, the expected positive linear relationship between PaCO2 and CBF may be absent on postnatal day one. On postnatal day three, and possibly day two, a PaCO2 threshold exists for this relationship, above which CBF becomes reactive to PaCO2. We speculate that the enhanced CBF response to PaCO2 above the threshold contributes to the reperfusion injury and partly explains the association between hypercapnia and peri/intraventricular haemorrhage.
INTRODUCTION Permissive hypercapnia has become a common lung protective strategy in ventilated preterm infants. However, the safety and the optimal range of carbon dioxide (CO2) values for permissive hypercapnia are not known (1–3). Alterations in arterial blood CO2 (PaCO2) can significantly affect cardiovascular function and cerebral haemodynamics (4). A rise in PaCO2 results in cerebral vasodilation and increases in cerebral blood flow (CBF), while a reduction in PaCO2 leads to cerebral vasoconstriction and decreases in CBF. It is notable that PaCO2 is the most potent acute regulator of CBF, as changes in PaCO2 induce proportionally greater changes in CBF than changes
in blood pressure with mean blood pressure (MBP) values outside the autoregulatory range for CBF (5). As a consequence, CBF autoregulation is attenuated when PaCO2 increases (6–8). These findings have clinical relevance, as the alterations in CBF associated with changes in CO2 might predispose the preterm infant’s vulnerable brain to peri/intraventricular haemorrhage (P/IVH) and/or periventricular leukomalacia (PVL). Indeed, an association has been shown between high and low PaCO2 and the
Key notes
Abbreviations BPD, bronchopulmonary dysplasia; CBF, cerebral blood flow; CFOE, cerebral fractional oxygen extraction; MBP, mean blood pressure; MCA-MV, middle cerebral artery mean flow velocity; P/IVH, peri/intraventricular haemorrhage; PaCO2, partial pressure of arterial carbon dioxide; PaO2, arterial oxygen tension; PVL, periventricular leukomalacia; ROP, retinopathy of prematurity.
e334
Hypercapnia and cerebral hypoperfusion–reperfusion are associated with peri/intraventricular haemorrhage. We found that after an initial absence during the first postnatal day, cerebral blood flow (CBF) reactivity to carbon dioxide (CO2) is re-established by the second or third postnatal day in very preterm neonates. The steep positive linear relationship between arterial CO2 and CBF above an arterial CO2 level of 51– 53 mmHg may explain the association between hypercapnia and peri/intraventricular haemorrhage.
©2014 Foundation Acta Pædiatrica. Published by John Wiley & Sons Ltd 2014 103, pp. e334–e339
Noori et al.
occurrence of P/IVH and PVL, respectively, in preterm infants during the transitional period (1,9,10). As for the association between PaCO2 and P/IVH, maximum PaCO2 during the first three postnatal days appears to be a dosedependent predictor of severe P/IVH (1). As early postnatal brain ischaemia followed by reperfusion is implicated in the pathogenesis of P/IVH (11–14), the haemodynamically driven increase in CBF during the reperfusion phase is likely to be accentuated if concomitant increases in PaCO2 occur. This, at least, partly explains why hypercapnia is associated with an increased risk of P/IVH during the first postnatal days (1,9). Therefore, we hypothesised that there might be a threshold, above which the impact of PaCO2 on CBF is exaggerated. To test this hypothesis in a prospective observational study, we investigated the relationship between CBF and PaCO2. We sought to identify the threshold in PaCO2, beyond which CBF is more responsive to changes in PaCO2 in haemodynamically stable very preterm infants during the first three postnatal days when most P/IVH occur.
METHODS The institutional review board at the Children’s Hospital of the University of Oklahoma Health Sciences Center approved the study. Informed parental consent was obtained. We included preterm infants of ≤30 weeks’ gestation and three or less days of age who were born in the Center and had an indwelling arterial catheter. We excluded patients with conditions associated with myocardial dysfunction and/or impaired regulation of systemic vascular resistance potentially affecting CBF. These conditions included hypotension, defined as MBP less than gestational age in weeks, vasopressor-inotrope administration, a 5-min Apgar score of less than four, congenital heart defects and/or multiple congenital anomalies. Because there is a good correlation between Doppler indices of flow and other measures of CBF (15,16), we used middle cerebral artery mean flow velocity (MCA-MV) as a surrogate for CBF (15,16). We measured MCA blood flow pattern by the pulse wave Doppler technique using the temporal approach with an angle of insonation of 50 mmHg supports our finding of a threshold for the CBF– CO2 reactivity (18). These observations might, at least in part, also explain the observed association between hypercapnia and the increased risk of P/IVH in very preterm neonates during the first postnatal days (1,9). However, as
©2014 Foundation Acta Pædiatrica. Published by John Wiley & Sons Ltd 2014 103, pp. e334–e339
e337
Carbon dioxide and cerebral blood flow
Noori et al.
we did not continuously follow the changes in PaCO2 and monitor for the occurrence of P/IVH, we cannot ascertain a temporal relationship between the increase in PaCO2 and CBF and the occurrence of P/IVH. Importantly, we found that exclusion of preterm neonates who developed P/IVH from the analysis did not affect the PaCO2 and MCA-MV relationship or the PaCO2 breakpoints. This finding indicates that the pattern of CBF response to PaCO2 exists irrespective of the development of P/IVH. The evolution of the PaCO2–CBF relationship from none during the first postnatal day to the expected positive linear relationship by the third postnatal day is an intriguing finding. To our knowledge, this phenomenon has not been previously described, although an attenuated response to changes in PaCO2 during the first postnatal day has been reported (19,20). In preterm infants ≤33 weeks gestation, Levene et al. (19) assessed the effect of a 1-kPa (7.5 mmHg) increase in PaCO2 on anterior cerebral artery MV and reported an increase in the magnitude of CBF–CO2 reactivity from postnatal day one to days 2–9. Pryds et al. (20) measured CBF at different PaCO2 levels and found markedly lower CBF reactivity to CO2 during the first postnatal day than the second (about 11% per kPa PaCO2 vs. 32% per kPa PaCO2, respectively). Tyszczuk et al. also reported a weak relationship between PaCO2 and CBF using nearinfrared spectroscopy to directly measure CBF in preterm neonates during the first postnatal day (21). The reason for the difference in the magnitude of the CBF–PaCO2 reactivity during the first postnatal day between our study and the previous studies (19–21) is unclear. It is notable that Kissack et al. (22) found no independent effect of PaCO2 on cerebral fractional oxygen extraction (CFOE; a surrogate for CBF) in preterm neonates during the first three postnatal days. This finding might be explained by CFOE being a less sensitive measure of the changes in CBF and/or by the differences in the patient population between this study (22) and the other studies (19–21) as well as the present study. The mechanisms involved in the reported absence or attenuated response of CBF to changes in PaCO2 during the immediate transitional period are not known. However, we speculate that the overwhelming vasoconstriction of the cerebral vessels of the very preterm neonate occurring immediately after delivery might be, at least in part, responsible for this observation. Factors leading to cerebral vasoconstriction in the very preterm neonate after birth are likely include the delivery-associated decreased systemic blood flow (12) and compensated shock along with the suspected nonvital organ assignment of the forebrain vasculature (17), as well as the relative, physiologic postnatal hyperoxia occurring immediately after delivery. We only found a weak positive linear relationship between the MBP and MCA-MV after adjusting for PaCO2. This finding, along with the lack of a relationship between the MBP and MCA-MV for PaCO2 values
REGULAR ARTICLE
Effect of carbon dioxide on cerebral blood flow velocity in preterm infants during postnatal transition Shahab Noori ([email protected])1, Michael Anderson2, Sadaf Soleymani1,3, Istvan Seri1 1.Division of Neonatology and the Center for Fetal and Neonatal Medicine, Department of Pediatrics, Children’s Hospital Los Angeles and the LAC+USC Medical Center, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA 2.Biostatistics and Epidemiology, University of Oklahoma Health Sciences Center, Oklahoma City, OK, USA 3.Department of Biomedical Engineering, Viterbi School of Engineering, University of Southern California, Los Angeles, CA, USA
Keywords Haemodynamics, Middle cerebral artery, Neonate, Ultrasound, Vascular Doppler Correspondence S Noori, MD, Children’s Hospital Los Angeles, 4650 Sunset Blvd, MS# 31, Los Angeles, CA 90027, USA. Tel: (323) 361 5939 | Fax: (323) 361 7927 | Email: [email protected] Received 25 November 2013; revised 18 February 2014; accepted 25 March 2014. DOI:10.1111/apa.12646
ABSTRACT Aim: High arterial carbon dioxide (PaCO2) and cerebral reperfusion are associated with peri/intraventricular haemorrhage. Our aim was to study the relationship between PaCO2 and cerebral blood flow (CBF) in preterm infants during postnatal transition. Methods: We prospectively studied ≤30 weeks’ gestation haemodynamically stable preterm infants during the first three postnatal days (n = 21; gestational age 25.8 1.4 weeks). We measured middle cerebral artery mean flow velocity (MCA-MV) as a surrogate for CBF at the time of blood gas analysis. Results: We obtained 78 PaCO2–MCA-MV data pairs. The expected positive linear relationship between PaCO2 and MCA-MV was absent on the first postnatal day, equivocal on the second and present on the third. Using piecewise bilinear regression models, we identified PaCO2 breakpoints at 52.7 and 51.0 mmHg for postnatal days two and three, respectively. Conclusion: In haemodynamically stable preterm neonates, the expected positive linear relationship between PaCO2 and CBF may be absent on postnatal day one. On postnatal day three, and possibly day two, a PaCO2 threshold exists for this relationship, above which CBF becomes reactive to PaCO2. We speculate that the enhanced CBF response to PaCO2 above the threshold contributes to the reperfusion injury and partly explains the association between hypercapnia and peri/intraventricular haemorrhage.
INTRODUCTION Permissive hypercapnia has become a common lung protective strategy in ventilated preterm infants. However, the safety and the optimal range of carbon dioxide (CO2) values for permissive hypercapnia are not known (1–3). Alterations in arterial blood CO2 (PaCO2) can significantly affect cardiovascular function and cerebral haemodynamics (4). A rise in PaCO2 results in cerebral vasodilation and increases in cerebral blood flow (CBF), while a reduction in PaCO2 leads to cerebral vasoconstriction and decreases in CBF. It is notable that PaCO2 is the most potent acute regulator of CBF, as changes in PaCO2 induce proportionally greater changes in CBF than changes
in blood pressure with mean blood pressure (MBP) values outside the autoregulatory range for CBF (5). As a consequence, CBF autoregulation is attenuated when PaCO2 increases (6–8). These findings have clinical relevance, as the alterations in CBF associated with changes in CO2 might predispose the preterm infant’s vulnerable brain to peri/intraventricular haemorrhage (P/IVH) and/or periventricular leukomalacia (PVL). Indeed, an association has been shown between high and low PaCO2 and the
Key notes
Abbreviations BPD, bronchopulmonary dysplasia; CBF, cerebral blood flow; CFOE, cerebral fractional oxygen extraction; MBP, mean blood pressure; MCA-MV, middle cerebral artery mean flow velocity; P/IVH, peri/intraventricular haemorrhage; PaCO2, partial pressure of arterial carbon dioxide; PaO2, arterial oxygen tension; PVL, periventricular leukomalacia; ROP, retinopathy of prematurity.
e334
Hypercapnia and cerebral hypoperfusion–reperfusion are associated with peri/intraventricular haemorrhage. We found that after an initial absence during the first postnatal day, cerebral blood flow (CBF) reactivity to carbon dioxide (CO2) is re-established by the second or third postnatal day in very preterm neonates. The steep positive linear relationship between arterial CO2 and CBF above an arterial CO2 level of 51– 53 mmHg may explain the association between hypercapnia and peri/intraventricular haemorrhage.
©2014 Foundation Acta Pædiatrica. Published by John Wiley & Sons Ltd 2014 103, pp. e334–e339
Noori et al.
occurrence of P/IVH and PVL, respectively, in preterm infants during the transitional period (1,9,10). As for the association between PaCO2 and P/IVH, maximum PaCO2 during the first three postnatal days appears to be a dosedependent predictor of severe P/IVH (1). As early postnatal brain ischaemia followed by reperfusion is implicated in the pathogenesis of P/IVH (11–14), the haemodynamically driven increase in CBF during the reperfusion phase is likely to be accentuated if concomitant increases in PaCO2 occur. This, at least, partly explains why hypercapnia is associated with an increased risk of P/IVH during the first postnatal days (1,9). Therefore, we hypothesised that there might be a threshold, above which the impact of PaCO2 on CBF is exaggerated. To test this hypothesis in a prospective observational study, we investigated the relationship between CBF and PaCO2. We sought to identify the threshold in PaCO2, beyond which CBF is more responsive to changes in PaCO2 in haemodynamically stable very preterm infants during the first three postnatal days when most P/IVH occur.
METHODS The institutional review board at the Children’s Hospital of the University of Oklahoma Health Sciences Center approved the study. Informed parental consent was obtained. We included preterm infants of ≤30 weeks’ gestation and three or less days of age who were born in the Center and had an indwelling arterial catheter. We excluded patients with conditions associated with myocardial dysfunction and/or impaired regulation of systemic vascular resistance potentially affecting CBF. These conditions included hypotension, defined as MBP less than gestational age in weeks, vasopressor-inotrope administration, a 5-min Apgar score of less than four, congenital heart defects and/or multiple congenital anomalies. Because there is a good correlation between Doppler indices of flow and other measures of CBF (15,16), we used middle cerebral artery mean flow velocity (MCA-MV) as a surrogate for CBF (15,16). We measured MCA blood flow pattern by the pulse wave Doppler technique using the temporal approach with an angle of insonation of 50 mmHg supports our finding of a threshold for the CBF– CO2 reactivity (18). These observations might, at least in part, also explain the observed association between hypercapnia and the increased risk of P/IVH in very preterm neonates during the first postnatal days (1,9). However, as
©2014 Foundation Acta Pædiatrica. Published by John Wiley & Sons Ltd 2014 103, pp. e334–e339
e337
Carbon dioxide and cerebral blood flow
Noori et al.
we did not continuously follow the changes in PaCO2 and monitor for the occurrence of P/IVH, we cannot ascertain a temporal relationship between the increase in PaCO2 and CBF and the occurrence of P/IVH. Importantly, we found that exclusion of preterm neonates who developed P/IVH from the analysis did not affect the PaCO2 and MCA-MV relationship or the PaCO2 breakpoints. This finding indicates that the pattern of CBF response to PaCO2 exists irrespective of the development of P/IVH. The evolution of the PaCO2–CBF relationship from none during the first postnatal day to the expected positive linear relationship by the third postnatal day is an intriguing finding. To our knowledge, this phenomenon has not been previously described, although an attenuated response to changes in PaCO2 during the first postnatal day has been reported (19,20). In preterm infants ≤33 weeks gestation, Levene et al. (19) assessed the effect of a 1-kPa (7.5 mmHg) increase in PaCO2 on anterior cerebral artery MV and reported an increase in the magnitude of CBF–CO2 reactivity from postnatal day one to days 2–9. Pryds et al. (20) measured CBF at different PaCO2 levels and found markedly lower CBF reactivity to CO2 during the first postnatal day than the second (about 11% per kPa PaCO2 vs. 32% per kPa PaCO2, respectively). Tyszczuk et al. also reported a weak relationship between PaCO2 and CBF using nearinfrared spectroscopy to directly measure CBF in preterm neonates during the first postnatal day (21). The reason for the difference in the magnitude of the CBF–PaCO2 reactivity during the first postnatal day between our study and the previous studies (19–21) is unclear. It is notable that Kissack et al. (22) found no independent effect of PaCO2 on cerebral fractional oxygen extraction (CFOE; a surrogate for CBF) in preterm neonates during the first three postnatal days. This finding might be explained by CFOE being a less sensitive measure of the changes in CBF and/or by the differences in the patient population between this study (22) and the other studies (19–21) as well as the present study. The mechanisms involved in the reported absence or attenuated response of CBF to changes in PaCO2 during the immediate transitional period are not known. However, we speculate that the overwhelming vasoconstriction of the cerebral vessels of the very preterm neonate occurring immediately after delivery might be, at least in part, responsible for this observation. Factors leading to cerebral vasoconstriction in the very preterm neonate after birth are likely include the delivery-associated decreased systemic blood flow (12) and compensated shock along with the suspected nonvital organ assignment of the forebrain vasculature (17), as well as the relative, physiologic postnatal hyperoxia occurring immediately after delivery. We only found a weak positive linear relationship between the MBP and MCA-MV after adjusting for PaCO2. This finding, along with the lack of a relationship between the MBP and MCA-MV for PaCO2 values
