Main

Current guidelines from the International Liaison Committee on Resuscitation breathe considerable uncertainty as to how much supplementary oxygen should be given during resuscitation of newborn asphyxiated infants (1). Previous recommendations were to be generous with oxygen, but recent guidelines from a number of countries, e.g. Australia, Canada, Finland, the Netherlands, Sweden, and the United Kingdom recommend initial ventilation with air, because of the results from several experimental (2–5) and clinical (6–13) studies indicating that resuscitation with 100% oxygen is harmful. In the clinical studies, the time of exposure to 100% oxygen was typically 5–7 min (8,13), whereas only one animal study (5) has investigated an exposure time to oxygen less than 15 min.

We speculated that very short times of exposure might not allow systemic hyperoxia to develop and so be harmless to the newborn infant, except for a possible negative effect on the lungs. In fact, the pulse oximetric saturation at 3 min after birth is usually below 80% (14,15) in the normal air-breathing infant, suggesting the presence of cardiac or pulmonary right-to-left shunts. One might expect that such shunts would delay the appearance of arterial hyperoxemia in the infant breathing pure oxygen. However, this has been studied neither in normal nor in asphyxiated subjects.

We used a sheep model of term intrauterine asphyxia with postnatal resuscitation, and hypothesized that hyperoxia of arterial blood and of brain tissue could be prevented by limiting the period of ventilation with 100% oxygen to 3 min. We also investigated whether the speed of circulatory recovery, as reflected by the heart rate (HR) and mean arterial blood pressure (MAP) responses, and the speed of recovery of brain oxygenation, as reflected by regional cerebral oxygen saturation (CrSO2), differed between oxygen- and air-resuscitation.

Work on the same animal model has shown that the mRNA expression of some proinflammatory cytokines in the brain is increased in asphyxiated lambs 90 min after exposure to 30 min of 100% oxygen breathing (16), and a secondary aim was to see whether already 3 min with pure oxygen could generate the same inflammatory response.

METHODS

The study was approved by the Animal Ethics Research Committee of Lund University. The animals were cared for and handled in accordance with European Guidelines for Use of Experimental Animals.

Animal preparation and cerebral oxygenation measurements.

Date-mated ewes were anesthetized and underwent cesarean section at 140–141 gestational days, when pulmonary surfactant function in the lamb is usually normal (17). Term is 145–150 d. The fetus received catheters in an axillary artery and a jugular vein, and was tracheotomized as described previously (16).

The scalp was shaved and an incision made to expose the right parietal bone. A hole of 3-mm diameter was drilled down to the dura mater, and a hollow bolt screwed into the hole. By using an introducer passed through the lumen of the bolt, a flexible microcatheter probe containing a polarographic cell (LICOX; GMS, Mielkendorf, Germany) was placed with its tip in the subcortex so that brain tissue Po2 (PbtO2) could be continuously measured. Alongside the PbtO2 probe, a thermocouple probe was placed through the same introducer, and both were connected to a LICOX CMP monitor for automatic temperature-corrected PbtO2 readings. Every 20 s, the PbtO2 and temperature signals were collected and stored in a personal computer using the LICOX for PC Software. The PbtO2 recording was later processed to find the individual's maximum (peak) value.

Measurement of CrSO2 by dual wavelength near-infrared spectroscopy, was performed with equipment for clinical use (INVOS® 5100C Cerebral/Somatic Oxymeter, Somanetics Corporation, Troy, MI), using a disposable self-adherent pediatric probe (Pediatric SomaSensor probe, <40 kg, model SPFB-USA, Somanetics Corporation) that was attached to the shaved scalp and secured in place by a fabric net.

Induction of asphyxia and resuscitation.

After clamping and cutting the umbilical cord, the lamb was removed from the womb, weighed, sedated, and placed in an open, heated incubator. A catheter was positioned in the abdominal aorta via the umbilical cord and used for recording MAP and HR on a Datex-Ohmeda AS/3 Compact Anesthesia Monitor (GE Health Care, Helsinki, Finland). Pressure-regulated volume control ventilation with a Servo 300 ventilator (Siemens-Elema, Solna, Sweden) was initiated exactly 10 min after cord occlusion. Tidal volume was 10 mL/kg, respiratory rate 40/min, I:E ratio 1:1, and end-expiratory pressure 4 cm H2O. The lambs were randomized to receive an inhaled oxygen fraction (FiO2) that was either 21% (n = 7) throughout or 100% during the first 3 (n = 6) or 30 min (n = 6) of ventilation. The ventilation with pure oxygen was followed by ventilation with air. One lamb of the 30-min oxygen group was subsequently excluded because of low blood hemoglobin (Hb) concentration; see Results. An observer, blinded to the FiO2, looked at the arterial pressure curve, to judge whether the circulation showed signs of spontaneous recovery after start of ventilation. If not, 10 μg/kg of adrenaline was injected i.v., and external cardiac massage given for 15 s. After the first half hour, respiratory rate was adjusted to achieve an arterial Pco2 (PaCO2) around 6 kPa.

Arterial sampling.

A 0.5 mL sample for blood-gas and Hb measurements was obtained from the axillary artery catheter shortly before occluding the umbilical cord (in utero), just before start of ventilation (end of asphyxia), and 0.75, 2.5, 5, 10, 15, 29.5, and 60 min after start of ventilation. The samples were analyzed on an ABL 700 blood-gas analyzer (Radiometer, Copenhagen, Denmark). Its settings were adjusted to ovine blood, according to a factory-installed algorithm.

Assessment of circulatory recovery and of recovery of cerebral oxygenation.

The time from start of ventilation until HR had increased to 150 bpm was measured. We also assessed the time from start of ventilation until CrSO2 reached 30%, i.e. a value just above the minimum displayed by the equipment, namely 15%. Because a new CrSO2 figure was presented every 6 s, and the HR and MAP every 10 s, interpolation was used to calculate the times to the nearest second.

In a separate analysis, individual HR and MAP curves were aligned, so that the respective points in time when HR reached 150 bpm coincided. The respective mean curves representing each group were then calculated.

Experiments in nonasphyxiated controls.

Subsequent to the asphyxia series, four nonrandom controls were ventilated with oxygen to assess cerebral oxygenation during hyperoxemia that was not preceded by asphyxia. Using an infant self-expanding resuscitator bag, manual ventilation with air started with the lamb in utero, a few seconds before clamping and cutting the umbilical cord. The lamb was transferred to the open incubator where mechanical ventilation with 100% oxygen was initiated as soon as an aortic catheter had been placed via the umbilical cord, for pressure monitoring. After 30 min, the inspired gas was changed to air. Axillary arterial samples for blood-gas and Hb measurements were taken in utero, and 0.75, 2.5, 5, 10, 15, 29.5, and 60 min after switching from air- to oxygen-ventilation. The preparation, ventilator settings, sedation, and measurements were otherwise the same as for the asphyxiated animals.

mRNA expression in the brain of proinflammatory cytokines IL-1β, IL-12, and IL-18.

After the 60 min arterial sample, the heart was arrested by an i.v. injection of potassium chloride. In 13 asphyxiated animals and three controls, the brain was immediately removed from the skull and split in the midline. In the left half, tissue sections were taken from the temporo-parietal cortex including sub-cortical white matter and from the thalamus, and immediately frozen on dry ice. The samples were stored at −80°C.

Total RNA was extracted from frozen tissue using Trizol® (GIBCO BRL, Invitrogen Corporation, Grand Island, NY) according to the manufacturer's instructions. Gene transcripts were quantified using real-time PCR on ABI PRISM® 7000 sequence detection system (Applied Biosystems, Foster City, CA) as described previously (10). The quantitative value of each sample was normalized to the corresponding value of beta-actin and results expressed as relative values.

Statistics.

Between-group differences among asphyxiated lambs were assessed by analysis of variance using Sigma Chemical Co.-Stat 3.5 software (Systat Software GmbH, Ekrath, Germany). In case of non-normal distribution, analysis of variance on ranks was used instead. The difference in peak PbtO2 between the group of asphyxiated lambs that was ventilated with oxygen for 30 min and the nonasphyxiated lambs was assessed with the t test.

RESULTS

One asphyxiated lamb, randomized to resuscitation with oxygen for 30 min, was excluded because of a low Hb: 60 g/L. Among the remaining 18 asphyxiated animals, the median (range) for Hb was 136 (119–157) g/L, with no difference between the three groups. Hb was 127 (95–151) g/L in the four nonasphyxiated animals.

Asphyxiated lambs.

Table 1 shows arterial blood gases and base excess (BE) by group. In utero pH (not shown in the Table) of all 18 subsequently asphyxiated animals was 7.26 (7.10–7.32). There was no significant difference between groups in respect of any of these measures.

Table 1 Arterial blood gases
Full size table

During the cord occlusion, median arterial PO2 (PaO2) decreased to 0.74 kPa, and a severe combined respiratory and metabolic acidemia developed, with a median PaCO2 of 17 kPa, and BE of −13 mM. pH decreased to 6.92 (6.81–6.99). PbtO2 became less than 0.1 kPa in all animals, and CrSO2 was displayed as the minimum value for the equipment, i.e. 15%, in all animals. MAP decreased to a nadir of 14 (7–20) mm Hg in 17 animals and to 32 mm Hg in one. HR decreased to 64 (30–96) bpm. There was no difference between the groups in any of these measures.

The circulation recovered spontaneously when ventilation was started, except in one lamb in the air-only group and in two in each of the two oxygen groups. These five lambs were successfully resuscitated with adrenaline and external cardiac massage. Time to reach a HR of 150 bpm was the same with oxygen-ventilation as with air (Table 2) and curves representing mean HR and MAP were similar in the three groups (Fig. 1). Median time to reach a CrSO2 of 30% was 2 min in those air-ventilated and ½–1 min in the two groups given 100% oxygen (Table 2). The difference between groups was not significant (p = 0.084).

Table 2 Times from start of ventilation until heart rate (HR) reached 150 bpm, and until cerebral regional oxygen saturation (CrSO2) reached 30% in asphyxiated lambs
Full size table
Figure 1
figure 1

Heart rate and arterial pressure response to resuscitation. (A) HR and MAP in a lamb that was resuscitated with oxygen for 3 min, and then with air ventilation. The arrow and horizontal bar indicate injection of adrenaline and subsequent cardiac compressions. All lambs had a similar abrupt increase in HR from less than 100/min to more than 200/min, concomitant with a likewise abrupt increase of arterial pressure, signaling that the resuscitation was successful. The time when this occurred varied between individuals, but not between groups (Table 2). (B) Mean curves for HR and MAP in the three groups of asphyxiated lambs. They were obtained by aligning curves from individual lambs, so that the times when HR reached 150/min coincided, and then computing the mean curve. This method was chosen in order that the mean curves should reflect the above-mentioned abrupt increase in HR and MAP.

Full size image

After 45 s of ventilation, PaO2 had increased slightly in most animals, but there was no arterial hyperoxemia (Table 1). Such was seen at 2 ½ min when PaO2 was over 50 kPa in three lambs ventilated with oxygen. In the samples taken at 5, 10, and 15 min, the median PaO2 was over 55 kPa in lambs allotted to prolonged oxygen breathing (Fig. 2C), and PbtO2 generally increased to peak values over 50 kPa in these (Fig. 2C, Table 2). Among lambs ventilated with pure oxygen for just 3 min, PbtO2 peaked at 46 kPa in one, and was 2.9–5.4 kPa in the other five (Fig. 2B). When oxygen-ventilation was discontinued, PbtO2 decreased to essentially the same levels as in the air-only group (Fig. 2).

Figure 2
figure 2

Brain tissue PO2 (PbtO2) and arterial PO2 (PaO2). Individual curves depicting PbtO2 (continuous lines) in lambs that were asphyxiated in utero and resuscitated after birth, with (A) air-ventilation (n = 7) or with 100% oxygen for (B) 3 min (n = 6) or (C) 30 min (n = 5), and then with air. Group medians for PaO2 (▪—▪) are also shown. The dot at −10 min represents the PaO2 in utero, before occluding the umbilical cord. (D) In a group that was not made asphyctic (n = 4), the PbtO2 and PaO2 during 30 min of oxygen ventilation and a subsequent 30 min of air ventilation are displayed.

Full size image

Nonasphyxiated controls.

At the intrauterine (baseline) stage, arterial oxygenation and PaCO2 were similar to those in the other groups (Table 1), but BE tended to be somewhat more negative. During 30 min of oxygen breathing, PaO2 levels were approximately the same as in those asphyxiated lambs that were ventilated equally long with oxygen (Fig. 2D), whereas peak PbtO2 was significantly less (Table 2).

Cytokines in cerebral tissue.

IL-1β, IL-12, and IL-18 (Table 3) did not vary significantly between groups.

Table 3 Cytokine mRNA activity in cerebral tissue
Full size table

DISCUSSION

The main finding of the present study is the extremely high PbtO2 of asphyxiated lambs, exposed to prolonged ventilation with 100% oxygen (Fig. 2C). The peak value was 56 (30–61) kPa, and such a high PbtO2 has been previously seen only under hyperbaric conditions (18). The finding is probably due to a combination of factors, namely high FiO2, postischemic hyperemia accentuated by metabolic and respiratory arterial acidemia, and reduced metabolism (19) in the postasphyctic brain. Surprisingly, Lyng et al. (20) found a much smaller peak PbtO2 in 1–2-d-old piglets that were asphyxiated by breathing an hypoxic mixture and then resuscitated with 30 min of oxygen-ventilation. In these, PbtO2 only increased to a maximum of approximately 5 kPa. The discrepancy could be the result of differences in equipment and species, different PO2 electrode placement in the brain, a PaCO2 following asphyxia which was probably less in their study, and the fact that they used a postnatal asphyxia model, rather than one involving the transition from intrauterine to postnatal life.

In the present study, we addressed the question whether arterial and cerebral hyperoxia could be avoided by reducing time of exposure to 100% oxygen. As it turned out, PaO2 was above 50 kPa already at 2.5 min in three lambs. Thus, the fluid-filled lungs did not protect the newly delivered lambs from hyperoxemia, except in the arterial sample taken 45 s after start of ventilation. In addition, one lamb in the 3-min oxygen group had an extremely high PbtO2 peak (Fig. 2B). Whether a similar rapid hyperoxygenation could occur in asphyxiated human neonates is not known. Vento et al. (8) found a PaO2 of approximately 17 kPa in arterial samples taken approximately 7 min after birth in asphyxiated infants breathing 100% oxygen. A suggestion that brief exposures could be toxic is given by clinical studies comparing oxygen and air ventilation for resuscitation after asphyxia. In these, an increased mortality (21) and signs of oxidative stress (13) were seen in infants given 100% oxygen during a mean time of only 7.5 ± 1.8 min. In addition, an association between neonatal exposure to 100% oxygen for less than 10 min and childhood leukemia has been found (10). In another study, neonatal exposure to supplemental oxygen for more than 3 min was associated with an increased risk of subsequent cancer, although a shorter exposure was not (11).

There was no significant increase in mRNA expression of proinflammatory cytokines after 30 min of 100% oxygen, in contrast to results in a previous paper (16). In that study, cerebral proinflammatory activity was evaluated at 2 h after initiation of resuscitation compared with 1 h in the present study. Sufficient time after hyperoxic exposure may be required for detection of increase in proinflammatory cytokine expression. In addition, the considerably smaller sample size in the present study limits detection of differences between groups.

The INVOS system, that was used to measure CrSO2, is adapted to the light absorption characteristics of human Hb, which are slightly different from those of ovine Hb (22). Nevertheless, the measurement was thought to provide an idea as to how quickly cerebral oxygenation was being restored during resuscitation, and we used the time taken for CrSO2 to reach 30% as an indicator. There was a tendency for this time to be shorter with oxygen ventilation than with air (Table 2), but the groups were not significantly different in this respect. The median time until HR had reached 150 bpm was the same with air ventilation as with oxygen (Table 2), and the MAP responses, too, were similar (Fig. 1). This is consistent with previous studies on the same asphyxia model (16) and with studies in asphyxiated 1–3-d-old piglets (23).

In conclusion, the findings in this model of fetal asphyxia followed by neonatal resuscitation indicate that ventilation with air will restore the circulation as fast as when pure oxygen is used, if ventilation is unobstructed and the lungs normal. Prolonged ventilation with 100% oxygen caused an extreme increase in brain tissue oxygen tension, which emphasizes the need to limit oxygen exposure in the asphyxiated newly born. However, reducing the time of 100% oxygen ventilation to 3 min did not consistently avert systemic hyperoxia.