Hypocarbia in Preterm Infants With Periventricular Leukomalacia: The Relation Between Hypocarbia and Mechanical Ventilation
Objective. The aim of this study was to elucidate the relationship between mechanical ventilation and hypocarbia in infants with periventricular leukomalacia (PVL).
Study Design. Matched pair analysis was conducted for 26 infants with PVL and 26 with normal development, who were born between 27 and 32 weeks' gestational age and required mechanical ventilation. The time-averaged carbon dioxide (CO2) index, Paco2, and pH were calculated every 24 hours for samples obtained from indwelling arterial catheters within the first 72 hours of life. The time-averaged respiratory rate of the ventilator (RR), peak inspiratory pressure (PIP), mean airway pressure (MAP), and ventilator index (VI) were also determined. The time-averaged total respiratory rate (TRR) was determined by observing the movement of the chest wall. The patients' characteristics, antenatal and neonatal variables, and electroencephalographic findings were also compared.
Results. The time-averaged CO2 index was larger, the time-averaged CO2 lower and the time-averaged pH higher in infants with PVL than in those with normal development on the third day of life. There was no significant difference in the time-averaged RR, PIP, MAP, or VI on any day. TRR was larger in the PVL group than in the control group on each day, but there was no significant difference. No significant difference was observed in the clinical characteristics or neonatal variables. Electroencephalographic abnormalities within 48 hours of life were more frequent in infants with PVL than in those with normal development.
Conclusion. Hypocarbia was associated with PVL because the time-averaged CO2 index was larger and the time-averaged Paco2 lower in infants with PVL than in those with normal development. However, the ventilator settings were similar among the infants with and without PVL.
Periventricular leukomalacia (PVL) is an important cause of cerebral palsy in preterm infants.1–4 Although PVL is regarded as a hypoxic-ischemic lesion of watershed areas on the basis of the results of pathologic studies,5–7 some recent studies suggested that other factors such as cytokines may be related to the development of white matter injury.8–11 Although many studies have been conducted, there have been few clinical risk factors for certainly predicting the occurrence of PVL so far. Among those studies, hypocarbia has attracted attention because some investigators agreed on the association between hypocarbia and the occurrence of PVL.12–17 Their association was explained by the speculation that cerebral vasoconstriction and diminished cerebral blood flow induced by hypocarbia, which are observed even in preterm infants,18 result in watershed hypoxic-ischemic lesions. Some authors suggested that hypocarbia is caused by excessive mechanical ventilation,14,15 but the relation between hypocarbia and PVL was not fully determined.
To elucidate the relationship between hypocarbia and PVL, 2 issues were addressed in this study: whether or not hypocarbia is related to the development of PVL, and whether or not mechanical ventilation is a cause of hypocarbia in patients with PVL.
PATIENTS AND METHODS
The study was conducted in the Department of Pediatrics, Anjo Kosei Hospital. Between 1992 and 1996, 248 infants between 27 and 32 weeks' gestational age were admitted to our hospital. Sixteen of them died during the neonatal period. Thirty-seven of the 232 surviving infants developed PVL. Ten infants with PVL did not require mechanical ventilation throughout the neonatal period. One infant developed PVL postnatally subsequent to severe necrotizing enterocolitis, which developed on the third day of life. The remaining 26 infants (PVL group) were selected as the subjects of this study to precisely assess the relation between mechanical ventilation and PVL. The control group was selected from among mechanically ventilated infants who achieved normal psychomotor development at 18 months of age. They consisted of 26 infants with gestational ages and birth weights matching those in the PVL group. An infant was chosen from among those who were born as near as possible to a corresponding infant with PVL. All the infants studied had no severe complications that can cause brain injury after 72 hours of life.
All the infants were subjected to conventional mechanical ventilation. Mechanical ventilation was started when it was difficult to maintain Pao2 ≥60 mm Hg despite a fraction of inspired oxygen (Fio2) ≥0.6, Paco2 ≤60 mm Hg, and pH ≥7.25. The goal of our ventilator settings was to maintain Pao2between 60 and 90 mm Hg, Paco2 between 30 and 60 mm Hg, and pH between 7.3 and 7.45, while minimization of the ventilator settings was always considered. When hypocarbia <30 mm Hg was observed, the ventilation rate and peak inspiratory pressure (PIP) were immediately reduced stepwise to 10 to 12 cm H2O and 15 to 20 times per minute, respectively.
In this study, day 1 was taken as the first 24 hours after birth, day 2 the second 24 hours of life, and day 3 the third 24 hours of life.
The following clinical characteristics were evaluated on the basis of the medical charts of the infants: gestational age, birth weight, 1- and 5-minute Apgar scores, chorioamnionitis (defined as maternal fever >38°C with or without uterine tenderness), premature rupture of the membranes (defined as rupture for >24 hours), mode of delivery, evidence of intrauterine growth retardation, initial pH value and base excess, and the use of bicarbonate within the first 24 hours of life.
Blood Gas Measures, Ventilator Settings, and Respiratory State
Blood gas values were measured for samples obtained via indwelling arterial catheters. The intervals between blood gas measurements were usually 4 to 6 hours during the first 3 days. Some additional analyses were performed when necessary. The numbers of samples in each patients (median [range]) were 7.5 [6–14], 5 [4–6], and 4 [4–7] in days 1, 2, and 3, respectively, in the PVL group. In the control group, those numbers were 7.5 [6–12], 5 [4–6], and 4 [4–6] in days, 1, 2, and 3, respectively. Mann-Whitney U tests did not reveal a significant difference in the numbers of samples between 2 groups.
Evaluation of the variables below was started when endotracheal intubation was performed, as insertion of the arterial catheter usually preceded it. Evaluation was discontinued when the endotracheal tube or arterial catheter was removed. In this study, the observation time was defined as the period when both mechanical ventilation and an arterial catheter were maintained on each day. Thus, the observation time on each day could be various among infants.
To quantify the cumulative effect of potentially adverse effects of hypocarbia, we calculated the area above the curve obtained with longitudinal data and threshold levels of Paco2according to the method described by Wiswell et al12 (Fig 1). The value obtained by this method is called the carbon dioxide (CO2) index in this study. We used threshold levels only for Paco2 values of 25 mm Hg, because Wiswell et al12 did not find that the cumulative degree of hypocarbia at the level of 20 or 15 mm Hg resulted in an increased likelihood of PVL. Because the observation time was various among infants, each CO2 index was divided by the observation time in hours to even its difference.
Every 24 hours we calculated the area under the curve obtained with longitudinal data for Paco2 (Fig 1) and pH, and the area under the stepwise curve for the respiratory rate of the ventilator (RR), PIP, mean airway pressure (MAP), and ventilator index (VI: peak inspiratory − end expiratory pressure × respiratory rate). The area under the curve for these values was divided by the observation time in hours. In this study, we call these values the time-averaged Paco2, pH, RR, PIP, MAP, and VI, respectively. The concentration of oxygen was diligently adjusted by the attending nursing staff to maintain arterial oxygen saturation (Sao2) between 93% and 98%. This did not allow us to measure the area under the curve. The minimal and maximal oxygen concentrations necessary to maintain appropriate blood gas values on each day were determined.
The total respiratory rate (TRR) was visually measured by the attending nursing staff every 4 hours for each infant. This measurement was made by observing the movement of the chest wall when an infant was at rest. It is difficult to completely distinguish the movement attributable to spontaneous respiration from that attributable to mechanical ventilation. Thus, TRR was considered to be the sum of the spontaneous and mechanical respiratory rates. We also calculated the area under the curve for the longitudinal data for TRR on each day. This area was divided by the observation time in hours. We call this value the time-averaged TRR.
We also evaluated the following: surfactant replacement, duration of mechanical ventilation, the date of successful extubation, duration of supplementary oxygen use, duration of pharmacologic treatment against apnea (intravenous doxapram and/or oral aminophylline), and presence of chronic lung disease that necessitated supplementary oxygen at 36 weeks' postconceptional age.
The medical charts of the infants were also reviewed for the presence or absence of the following characteristics: hypotension <30 mm Hg lasting for >30 minutes, bradycardia <90 beats per minute lasting for >30 minutes, oliguria (defined as an urine output <1 mL/kg/hour), pharmacologic treatment for patent ductus arteriosus, necrotizing enterocolitis diagnosed clinically, hypoglycemia (defined as a serum glucose level <30 mg/dL), hyperkalemia (defined as a serum potassium level >7.0 mEq/l), and use of steroids during the first 2 weeks of life.
Neuroimaging, Electroencephalography (EEG), and Follow-Up
Cranial ultrasonography was routinely performed at least twice within first week of life, and thereafter once to 3 times a week in all infants. EEG was also routinely recorded in all infants within the first 72 hours of life. Magnetic resonance imaging (MRI) was performed routinely at 40 weeks' postconceptional age. Additional MRI was performed during late infancy through early childhood when an infant revealed delayed development or neurologic abnormalities. Psychomotor development was examined every 3 months after discharge by 3 of the authors (A.O., F.H., T.K.) at least until 2 years of corrected age. Both motor and cognitive impairment were evaluated based on a complete neurologic examination and Tsumori-Inage developmental quotient at 18 months' corrected age.
The diagnosis of PVL was made on the basis of the results of ultrasonography and MRI. Ultrasonography during the neonatal period demonstrated a cystic change in the deep white matter in 14 infants between 10 and 27 days of age, and prolonged periventricular high echo density in 12 infants. Cystic change or high echo density were bilateral in all infants. No infants had evidence of intraventricular hemorrhage. MRI during late infancy or early childhood revealed ventriculomegaly with an irregular margin accompanied by a periventricular abnormal high intensity area on T2-weighted imaging in all 26 infants. All infants had spastic diplegia or quadriplegia. On the other hand, no abnormal ultrasonographic or MRI findings were obtained in any of the 26 infants in the control group. All infants had achieved normal motor and mental development (developmental quotient >90) at >2 years follow-up.
EEG was performed polygraphically using a bipolar montage with 8 surface electrodes (AF3, AF4, C3, C4, O1, O2, T3, and T4) according to the 10 to 20 international system, combined with electro-oculography, electrocardiography, and respiratory movement measurement.19–23 The initial EEGs were recorded mostly within the first 48 hours of life. All EEGs were evaluated by 3 well-trained neonatal neurologists (F.H., A.O., and T.K.), who were unaware of the ultrasonographic findings.
Our previous study revealed chronologic EEG changes in infants with PVL.22,23 In the acute stage of brain injury, EEG shows varying degrees of depression, ie, acute-stage EEG abnormalities. Acute-stage EEG abnormalities gradually improve with time and are replaced by recovery or chronic-stage EEG abnormalities. Thus, the presence of acute-stage EEG abnormalities is considered to indicate that a brain insult had occurred shortly before the EEG recording.
Statistical analysis between the 2 groups was performed by means of the unpaired t test for quantitative variables and Fisher's exact test for qualitative variables. The following variables were analyzed with the Mann-Whitney U test; Apgar score, time-averaged CO2 index, dose of bicarbonate within the first 24 hours, minimal and maximal appropriate oxygen concentrations, duration of mechanical ventilation, date of extubation, duration of oxygen use, and duration of treatment for apnea. Statistical significance was accepted at the level of P < .05.
Gestational ages and birth weights were well-matched. The Apgar scores at 5 minutes were slightly lower in the PVL group than in the control group, but the difference was not statistically significant. Premature rupture of the membranes and cesarean section were frequent in both groups. Chorioamonionitis was observed in 5 infants in each groups. A significant difference was not observed in any clinical characteristics of the infants (Table 1).
Neonatal adverse events, except for treatment for patent ductus arteriosus, were rare in both groups. Mild necrotizing enterocolitis only developed in 1 infant in the control group at 28 days of age. Treatment for patent ductus arteriosus was common in both groups. Steroids were not administered in any infants. There was no significant difference in the incidence of any neonatal variables between the PVL and control groups (Table 2).
Acidosis was common in both groups on day 1, but was corrected within 1 or 2 hours after mechanical ventilation was started in all infants. Acidosis was observed in 1 infant in the PVL group and 2 infants in the control group on days 2 and 3, but it was quite transient. Hypoxia was common in both groups on day 1, but improved immediately after mechanical ventilation, and surfactant replacement, when necessary, was performed. Hypoxia was observed in 4 infants in the PVL group and 5 infants in the control group on days 2 and 3, but it was also quite transient.
Blood Gas Values
The numbers of mechanically ventilated infants in the PVL group were 25, 24, and 21 on days 1, 2, and 3, respectively. The arterial line was removed from 1 mechanically ventilated infant on day 2 and in another on day 3. Thus, evaluation of the following items was possible in 25 infants on day 1, 23 infants on day 2, and 19 infants on day 3. The number of mechanically ventilated infants in the control group was 25, 25, and 21 on days 1, 2, and 3, respectively. The arterial lines were maintained in all infants.
The median time-averaged CO2 index in the PVL group was 0.55 (25th–75th percentile, 0–2.75) on day 1, 0 (0–1.97) on day 2, and 1.24 (0.08–4.39) on day 3 (Fig 2). The median time-averaged CO2 index in the control group was 0.03 (0–0.60) on day 1, 0 (0–1.32) on day 2, and 0 (0–0.86) on day 3 (Fig 2). The Mann-Whitney U test revealed a significant difference on day 3 (P = .013), but not on days 1 or 2. The time-averaged Paco2 and pH on each day are shown on Fig 3. The unpaired ttest revealed a significant difference in the time-averaged Paco2 on day 3 (P = .004), but not on days 1 or 2. A significant difference in the time-averaged pH between 2 groups was also observed on day 3 (P = .003), but not on days 1 or 2.
Ventilator Settings and Respiratory States of Infants
The time-averaged RR, PIP, MAP, and VI are shown on Fig 4. There was no significant difference in the time-averaged RR, PIP, MAP, or VI on any day between the 2 groups. The ventilator settings were quite similar in the 2 groups even on days 3, when the time-averaged CO2 index, Paco2, and pH were significantly different between them.
The time-averaged TRR is shown on Fig 5. TRR was larger in the PVL group than in the control group on each day, but there was no significant difference. TRR on day 3 was 43.5 ± 1.4 (mean ± standard error of the mean) in the PVL group and 38.5 ± 2.0 in the control group. This difference was not statistically significant (P = .053).
The percentage of infants who underwent surfactant replacement was not different between the PVL and the control groups (Table 3). No significant difference was observed in the minimal or maximal appropriate oxygen concentration on each day, duration of mechanical ventilation, date of successful extubation, duration of supplementary oxygen use, duration of pharmacologic treatment for apnea or occurrence of chronic lung disease (Table 3).
EEGs were recorded within 48 hours of age in 22 infants in the PVL group and in 22 infants in the control group. Twenty (91%) of 22 infants in the PVL group had acute-stage EEG abnormalities, whereas 7 (32%) of 22 infants in the control group did so. Acute-stage EEG abnormalities were more frequent in infants in the PVL group than in the control group (P < .01).
As to the infants who required mechanical ventilation on day 3, EEGs had already been recorded in 20 of 21 infants in the PVL group and in all 21 infants in the control group. Acute-stage EEG abnormalities were present in all but 1 infant in the PVL group, whereas they were observed in only 7 infants (33%) in the control group (P < .01).
There have been many studies on the clinical risk factors for the development of PVL.12–17,24–27 The relation of hypocarbia to the occurrence of PVL or cerebral palsy in preterm infants has been well-described and a considerable number of investigators have agreed on their association.12–17,28Hypocarbia during artificial ventilation therapy was reported to be associated with an increased risk of PVL whether a conventional intermittent-positive pressure ventilator or a high-frequency jet ventilator was used.12,13 In our study, the time-averaged CO2 index was larger and the time-averaged Paco2 was lower in the PVL group than in the control group on day 3. These results indicate not only that the Paco2 level is generally low, but also that profound hypocarbia is common in infants with PVL. Hypocarbia was found to be associated with the occurrence of PVL, in agreement with the previous studies. Some reports suggested excessive ventilator settings were the cause of hypocarbia in patients with PVL.14,15However, the role of mechanical ventilation in hypocarbia has not yet been established because ventilator settings, such as the PIP, MAP, or rate of ventilation, were not sufficiently mentioned in those reports.
We selected matched pair analysis to elucidate the relation between ventilator settings and hypocarbia in infants with PVL, because we considered that the physiologic state was quite similar in infants with and without PVL to exclude the possible effects of other factors. If critically ill infants with poor outcomes were included, the adverse effects of the critical conditions themselves may exceed those attributable to hypocarbia. Conversely, if a large number of well infants with favorable outcomes were included, the morbidity of infants without PVL is likely to be underestimated. As a result of careful selection of infants, those included in this study were well-matched not only in gestational age and birth weight but also in the physiologic state. The Apgar scores, initial blood gas values, and dose of bicarbonate were not different between infants with and without PVL. Severe early neonatal complications were uncommon in both groups, although treatment for patent ductus arteriosus was common. Therefore, the development of PVL in our study is hardly attributable to these postnatal complications. We consider that these conditions allow us to precisely discuss the relation between ventilator settings and hypocarbia.
It is remarkable that TRR on day 3 was larger in the PVL group than in the control group although the difference was not statistically significant. This implies that the rate of spontaneous respiration in infants with PVL may be greater than that in infants without PVL, because the rate of mechanical ventilation was similar in them. There may be the possibility that spontaneous respiration is relatively excessive in infants with PVL. However, it is very hard to determine the adequacy of spontaneous respiration or its effect on the Paco2 level. The difference in TRR between infants with and without PVL was only 5 times per minute. It is uncertain whether or not such a small difference can cause a significant difference in the Paco2 level. In addition, measurement of spontaneous respiration itself is also difficult, because its rate and depth are not always regular. More precise evaluation is necessary to clarify the relation of spontaneous respiration to hypocarbia and the development of PVL.
The ventilator settings were quite similar among infants with and without PVL on each day, although hypocarbia was surely present in infants with PVL on day 3. There can be 2 explanations for this result. First, the ventilator settings in infants with PVL were relatively excessive, although unintentional. The same ventilator settings in infants with less severe lung disease could lead to hyperventilation. In this study, pulmonary function data such as tidal volume and compliance were not investigated. There is a possibility that infants with PVL had better lung function and that their ventilator settings became relatively excessive. The infants with PVL may have represented a subset of infants who should have been weaned more rapidly. Second, mechanical ventilation is not a main cause of hypocarbia in infants with PVL. Hypocarbia in infants with PVL was observed only on day 3 when the ventilator settings were very low. If mechanical ventilation was a main cause of hypocarbia, it is likely that hypocarbia had occurred earlier when high ventilator settings were required. Although the appropriateness of ventilator settings is difficult to determine, the ventilator settings for our infants on day 3 were very similar to those described for a study on permissive hypercapnia in preterm infants.29 This suggests that the ventilator settings for our infants are unlikely to be excessive. Larger TRR in infants with PVL will be noteworthy. If infants with PVL had better lung function, ie, higher compliance and larger tidal volume than those without PVL, spontaneous respiration rate in the former infants should be lower on condition that respiratory coordination is similar among those with and without PVL. However, our study showed the opposite result. In infants with PVL, TRR was larger despite lower Paco2and larger CO2 index. This suggests that respiratory coordination may be deranged in infants with PVL.
There is the essential question of whether hypocarbia is a cause or a result of PVL, although hypocarbia was regarded as a risk factor for later development of PVL in many reports including this one. Some authors explained that hypocarbia could be a cause of PVL according to the hypothesis that hypocarbia leads to a substantial decrease in cerebral blood flow resulting in white matter injury. Some authors reported that the reactivity of cerebral blood flow to changes in Paco2 was maintained in normal preterm infants,30,31 but was attenuated in infants with brain injury.32,33 However, there have been no reports directly indicating decreased cerebral blood flow in human infants during the period when hypocarbia was observed. Although some experimental studies revealed that cerebral blood flow was decreased in association with hypocarbia,34,35 the hypocarbia in these studies was induced by hyperventilation, which is not usually applied to human preterm infants. In addition, these studies revealed changes in cerebral blood flow but did not provide evidence of brain injury. Some experimental studies have demonstrated a protective effect of carbon dioxide.36,37 In those studies, brain damage induced by the unilateral carotid artery ligation model was less severe in hypercapnic rats than in hypocapnic or normocapnic ones. This indicates that hypercapnia can be protective when hypoxia-ischemia has already occurred and persists, but does not show that brain damage can be caused by hypocarbia in formerly intact participants. Hypoxia-ischemia responsible for brain damage was not likely to have existed during the hypocarbia in our infants with PVL because their physical condition had been maintained as well as that in infants without PVL. The hypothesis that hypocarbia is a cause of PVL lacks firm evidence. On the other hand, we consider that hypocarbia can be a result or a sequence of PVL. It is remarkable that EEG abnormalities were frequent in infants with PVL from very early in the neonatal period in our study. Acute-stage EEG abnormalities within 48 hours of life were significantly more frequent in infants with PVL. This suggests that the brains of infants who later developed PVL had already been injured before EEG was performed. The timing of brain injury is presumed, at the latest, to be immediately after birth, because the physiologic states of the studied infants were quite stable even in those with PVL from very early in the neonatal period. This result was consistent with our previous study, which revealed that the presumed timing of brain injury was perinatal or prenatal in 20 of 26 infants with PVL.22 The larger TRR in infants with PVL may also be attributable to preceding brain injury or a hypoxic-ischemic event. Gleed and Mortola38 reported that newborn rats exhibited hyperapnea when they were exposed to chronic intrauterine hypoxia induced by a simulated high altitude. The larger TRR in infants may be explained by preexisting hypoxic stress. However, our speculation is still preliminary. Additional investigation is necessary to establish the relation between EEG abnormalities and the development of PVL.
Hypocarbia was found to be associated with PVL on the basis of that the time-averaged CO2 index was larger and the time-averaged Paco2 was lower in the PVL group than in the control group. However, the ventilator settings were similar in the infants with and without PVL. Larger TRR in infants with PVL suggests that spontaneous respiration may be excessive in them, although the difference did not reach statistical significance. To clarify the relation between hypocarbia and PVL, some prospective studies on a larger number of infants are necessary.
- Received February 23, 2000.
- Accepted July 31, 2000.
Reprint requests to (A.O.) Department of Pediatrics, Nagoya University School of Medicine, 65 Tsurumai-cho, Showa-ku, Nagoya, Aichi 466-8550, Japan. E-mail:
- PVL =
- periventricular leukomalacia •
- Fio2 =
- fraction of inspired oxygen •
- PIP =
- peak inspiratory pressure •
- CO2 =
- carbon dioxide •
- RR =
- respiratory rate of the ventilator •
- MAP =
- mean airway pressure •
- VI =
- ventilator index •
- Sao2 =
- arterial oxygen saturation •
- TRR =
- total respiratory rate •
- EEG =
- electroencephalography •
- MRI =
- magnetic resonance imaging
- Dubowitz LMS,
- Bydder GM,
- Mushin J
- Wiswell TE,
- Graziani LJ,
- Kornhauser MS,
- et al.
- Graziani LJ,
- Spitzer AR,
- Mitchell DG,
- et al.
- Trounce JQ,
- Shaw DE,
- Levene MI,
- Rutter N
- Leahy FAN,
- Cates D,
- MacCallum M,
- Rigatto H
- Watanabe K. The neonatal electroencephalogram and sleep-cycle patterns. In: Eyre JA, ed. The Neurophysiological Examination of the Newborn Infant. New York, NY: Mac Keith Press; 1992:11–47
- Hayakawa F,
- Okumura A,
- Kato T,
- Kuno K,
- Watanabe K
- Perlman JM,
- Risser R,
- Broyles RS
- Mariani G,
- Cifuentes J,
- Carlo WA
- Wilson DF,
- Pastusko A,
- DiGiacomo JE,
- Pawlowski M,
- Schneiderman R,
- Delivoria-Papadopoulos M
- Vannucci RC,
- Towfighi J,
- Heitjan DF,
- Brucklacher RM
- Gleed RD,
- Mortola JP
- Copyright © 2001 American Academy of Pediatrics