Long-term histological outcome after post-hypoxic treatment with 100% or 40% oxygen in a model of perinatal hypoxic-ischemic brain injury

Abstract

Hypoxic newborns have traditionally been given supplemental oxygen, and until recently, guidelines for neonatal resuscitation recommended that 100% oxygen be used. Exposure to 100% oxygen after hypoxic injury, however, may exacerbate oxidative stress. The current study evaluated the effect of exposure to 100%, 40% or 21% oxygen after neonatal hypoxic-ischemic injury on the severity of brain injury after long term survival. The severity of histological brain injury was not different in animals exposed to 100% oxygen versus room air. Male animals treated with 40% oxygen post-hypoxia had the lowest mean total histology scores, but this was not statistically significant due to the large variation in injury within each treatment group.

These results support the growing number of studies in human infants and experimental animals that show no benefit of 100% oxygen over room air for neonatal resuscitation. Our results suggest that post-hypoxia treatment with 40% oxygen may be beneficial, particularly in males. Further studies of the effects of different concentrations of oxygen on brain injury are warranted, and should have sufficient power to examine sex differences.

Introduction

Perinatal hypoxic-ischemic brain injury remains an important cause of neonatal morbidity and mortality, particularly in premature infants. It has been estimated that per 1000 births, 1−2 infants have suffered perinatal hypoxia/ischemia (H/I) severe enough to cause cerebral palsy (Kuban et al., 1994; Stanley, 1992); uncounted others have milder injuries that may result in mental retardation, seizures, learning disorders, and behavioral disorders. Although advances in obstetrical and neonatal care have improved outcomes for both term and preterm infants, greater numbers of markedly preterm infants are now surviving, increasing the numbers of infants who are at risk for the sequelae of H/I (Murphy et al., 1995; Piecuch et al., 1997; Stanley, 1992).

Hypoxic newborns, whether due to failure to breathe adequately at birth or later hypoxic or ischemic events, have traditionally been given oxygen supplementation, and until recently, guidelines for neonatal resuscitation recommended that 100% oxygen be used (Kattwinkel et al., 1999). Exposure to 100% oxygen, however, may exacerbate oxidative stress that has occurred due to hypoxic injury. Recent animal studies and clinical trials have shown that resuscitation of newborns with room air may be more beneficial than resuscitation with 100% oxygen (reviewed in (Davis et al., 2004; Niermeyer et al., 2004; Saugstad et al., 2005). The most recent treatment recommendations for neonatal resuscitation emphasize the importance of adequate ventilation and do not recommend a specific concentration of oxygen (ILCOR, 2006).

There are few experimental studies of the effects of hyperoxia on the cellular and molecular events initiated by hypoxia in the perinatal brain. In piglet models of neonatal resuscitation, administration of 100% oxygen after hypoxia was shown to increase extracellular glycerol, matrix metalloproteinase levels (Munkeby et al., 2004); increase expression of the immediate-early gene Egr-1 mRNA and cell death in the medulla (Tyree et al., 2006); to result in persistent decrease in Na⁺,K⁺-ATPase activity (Goplerud et al., 1995); and result in a poorer early neurological outcome (Temesvari et al., 2001) compared to animals resuscitated with room air. Felderhoff-Mueser (Felderhoff-Mueser et al., 2005) reported apoptotic cell death in 7 day old rats exposed to as little as 2 hours of 80% oxygen, without a preceding hypoxic insult.

Our preliminary experiments using a rat model of perinatal H/I showed that pups exposed to 100% oxygen for 2 hours after H/I had less histological injury than those exposed to room air after H/I when examined at 12 hours after hypoxia. At 6 weeks after hypoxia, however, the 100% oxygen treated group had greater injury than the room air group (Noppens K, et al. XXIInd International Symposium on Cerebral Blood Flow, Metabolism and Function, 2005, Abstract 493). When analyzing this data, however, there was concern that the room air animals were relatively hypothermic in the post-hypoxic period compared to the 100% oxygen treated pups, and that this hypothermia may have been neuroprotective for pups experiencing hypoxia-ischemia (Bona et al., 1998; Thoresen et al., 1996). The current study was designed to test the hypothesis that exposure to 100% oxygen after H/I at constant temperature would increase the severity of brain injury after long term survival. Since many neonatologists are now using variable concentrations of oxygen for neonatal resuscitation in clinical practice (Leone et al., 2006; O'Donnell et al., 2004), we included an additional group exposed to 40% oxygen after H/I. We believe that this is the first experimental study of perinatal H/I brain injury to test the effect of treating animals post-hypoxia with a level of oxygen intermediate between 100% and room air.

Experimental Procedures

Hypoxia-Ischemia

All procedures involving animals were conducted according to criteria approved by the Institutional Animal Care and Use Committee at Oregon Health & Science University. Timed-pregnant Wistar rats were purchased from a commercial breeder (Harlan), housed in individual cages, and fed a standard laboratory chow ad libidum. Pups were delivered spontaneously and maintained with their dam; the day of birth was defined as postnatal (P) day 0. Litters were culled to 10 pups on P1 or P2. On P7, H/I was induced as described by Rice (Rice et al., 1981) with some modifications (Grafe, 1994). Pups were anesthetized with isoflurane and the left common carotid artery was permanently ligated by electrocautery. The surgical procedure was performed with pups kept warm on a Deltaphase Isothermal pad (Braintree Scientific). Pups were returned to their mother for 2−4 hours, then exposed to 8% oxygen/balance nitrogen in a humidified chamber with an ambient floor-level temperature of 35−36°C for 90 minutes. Female and male pups from 10 litters were assigned to treatment groups so that each treatment group contained pups from at least 4 different litters. In hypoxia only (H) groups, the pups remained in the temperature-controlled chamber for an additional 120 minutes, but with room air. In hypoxia-hyperoxia groups (HH), the chamber gas was changed to either100% (HH100) or 40% (HH40) oxygen for 120 minutes. After initially saturating the chamber with the desired gas, flow rates were decreased to the minimal rate needed to maintain the appropriate oxygen concentrations. Temperature and oxygen concentration were continually monitored while pups were in the chamber. Normal controls pups were removed from the dam for the same periods of time as the ligated animals and kept warm in room air. Pups were weaned at 21 days of age and housed with littermates of the same sex until euthanization at 6 weeks after hypoxia.

To directly determine the effects of the post-hypoxia environment on body temperature, rectal temperature was measured in a separate group of animals prior to hypoxia, at 30 minute intervals during the 90 minute hypoxia period and 2 hour post-hypoxia period, and at 1 hr intervals for another 2 hours. One group was returned to the mother at the end of the 90 minute hypoxia period (Nest), the other group was returned to the mother after 90 minutes hypoxia followed by 120 minutes of 100% oxygen in the 36°C chamber (HH100). Animals from 2 litters were randomly assigned to Nest or HH100 groups. These animals were euthanized at the end of the temperature measurement period and the brains were not examined histologically.

Morphological Analysis

The rats were euthanized with an intraperitoneal injection of 120 mg/kg pentobarbital (Nembutal) and then perfused with saline followed by cold 4% paraformaldehyde. After removal from the skull, the brains were coded with randomly assigned numbers. The forebrains were blocked in the coronal plane, the right side notched, and the brains were dehydrated, cleared, and embedded in paraffin. Sections were cut at a thickness of 6 μm. Two coronal sections at each of two levels of the forebrain (corresponding to approximately levels 18 and 28 of (Swanson, 2004) were stained with hematoxylin and eosin and examined to evaluate the extent of injury. A defined, full-thickness region of parietal cortex divided into 6 subregions, hippocampal regions CA1, CA3 and dentate gyrus, thalamus, and caudate-putamen, were examined bilaterally for injury (Figure 1). Sections were evaluated for neuron loss, disruption of the cytoarchitecture, gliosis and cavitation. Injury was scored in each region according to the following semi-quantitative scale: 0, normal; 1, mild injury; 2, moderate injury; 3, severe injury; and 4, complete cavitary infarct. In this model, mild injury often consists of scattered dead neurons or small, patchy clusters of dead neurons, making this type of injury score more useful than calculating “infarct volume.” Scores for the 11 regions on each side were totaled to give a total histology score between 0 and 44. All sections were evaluated without knowledge of treatment group.

Figure 1.

Brain regions examined for histological injury. C = neocortex; CP = caudate-putamen, H = hippocampus (subregions CA1, CA3, d = dentate gyrus), T = thalamus. The cortical region demarcated by the long lines was subdivided into six sub-regions (short lines). The notch marks the hemisphere contralateral to carotid occlusion.

Statistical Analysis

Statistical analyses were performed using GraphPad Prism version 4.00 for Macintosh, GraphPad Software, San Diego California USA or SigmaStat, Systat Software, San Jose California. Body weights were compared between groups using 2-way analysis of variance (ANOVA) with Bonferroni post-hoc comparions. Correlations between brain and body weights and histology scores were determined using Pearson's correlation coefficient. Histology scores were analyzed using Kruskal-Wallis analysis of variance on ranks. Histology scores were also analyzed with 2-way ANOVA, to see if there were interactions between sex and treatment group, while acknowledging that these data are categorical do not have a normal distribution. Friedman's 2-way ANOVA on ranks could not be performed due to unequal group sizes. Rectal temperatures were compared between groups and over time using 2-way repeated measures ANOVA with Bonferroni post-hoc comparisons.

Results

Hypoxia versus Hypoxia-Hyperoxia comparisons: Pre-hypoxia body weights did not differ between treatment groups or sex. Three animals died during the hypoxia period (1 female, 2 male) and two animals died between 12 and 17 days of age (1 H female, 1 HH40 male). At 7 weeks of age, males were heavier than females for all post-hypoxia treatment groups, and HH40 females weighed less than normal females and H females (2-way ANOVA with Bonferoni post-test p < 0.05) (Table 1). Two-way ANOVA of brain weights showed an effect of both treatment (p = 0.016) and sex (p < 0.0001). Bonferoni post tests showed that both HH100 and HH40 females had lower brain weights than normal females. The gross appearance of the brains in animals exposed to hypoxia ranged from normal to brains with large cystic infarcts (not shown). The presence of large cystic infarcts did not correlate with animal size or general behavior. There were no gross motor deficits in any animals. A single animal had identifiable external abnormalities (very low body weight, abnormally long and misaligned teeth; HH40 female).

Table 1.

Body and brain weights in grams at age 7 weeks, means +/− standard deviation; all surviving animals included. Body and brain weights of males were heavier than females within each treatment group. Two-way ANOVA of body weights showed effects of both treatment and sex, with Bonferoni post-test p < 0.05 (*) compared to normal control and room air groups. Two-way ANOVA of brain weights showed effects of both treatment and sex, with Bonferroni post-test p < 0.05 (#) compared to normal female controls.

		Post-hypoxia treatment group
	Normal controls	Room air	100% oxygen	40% oxygen
Female Body weight	188.2 +/− 7.05 (n = 5)	182.2 +/− 17.62 (n = 10)	177.6 +/− 8.66 (n = 14)	164.0 +/−0 19.50* (n = 11)
Male Body weight	257.5 +/− 17.02 (n = 4)	240.1 +/− 13.25 (n = 11)	245.4 +/− 16.07 (n = 11)	241.0 +/− 17.0 (n = 10)
Female Brain weight	1.81 +/− 0.08	1.64 +/− 0.17	1.59 +/− 0.11#	1.58 +/− 0.19#
Male Brain weight	1.89 +/− 0.11	1.79 +/− 0.15	1.80 +/− 0.21	1.83 +/− 0.11

Mean histology scores after 6 weeks survival are shown in Figure 2a. The number of animals per group is the same as reported in Table 1. Male animals treated with 40% oxygen after H/I had the lowest mean total histology scores, but Kruskal-Wallis analysis of variance on ranks showed no significant differences between groups. Two-way ANOVA showed a significant effect of sex (p = 0.023), but post-hoc comparisons were not significant due to the large standard deviations and relatively small number of animals in each group. The scatter plot of total histology scores in Figure 2b (lines at medians) shows that each group contained animals with no histological evidence of injury, accounting for the large standard deviations and the lack of statistical significance among groups. When males and females were combined in each treatment group, there were also no significant differences.

Figure 2.

A. Mean total histology scores with standard deviations; n = 10−14 per group. Kruskal-Wallis analysis of variance on ranks showed no significant differences between groups. Two-way ANOVA showed effect of sex (p = 0.023). Post-hoc comparisons were not significant, due to the large standard deviations. B. Scatter plot of histology scores (lines = medians) shows that each group contains animals with no histological evidence of injury. F = female, M = male, H = post-hypoxia room air (21% oxygen), HH100 = post-hypoxia 100% oxygen, HH40 = post-hypoxia 40% oxygen.

Body weight did not correlate with histology scores in most groups (the HH100 male body weight inversely correlated with the histology scores, p = 0.04; all other groups were not significantly different). Brain weights inversely correlated with histology scores for all groups (p values from 0.017 to <0.0001; data not shown).

Temperature analysis: Figure 3 shows rectal temperature over time in animals returned to their mother at the end of hypoxia (Nest; n = 6 at 0−210 minutes, n = 4 at 270 and 330 minutes) compared to animals exposed to 100% oxygen for an additional 120 minutes in the chamber (HH100; n = 5 at each time point). While in the hypoxia chamber, body temperature rose from the pre-hypoxia temperature of approximately 34°C to approximately 37°C. After 210 minutes total in the chamber, the HH100 group had a mean rectal temperature of 37.42 °C (range 36.0−38.7). In both groups body temperature decreased when the pups were initially returned to their mothers (Nest group 31.0−35.5, mean 33.78 at 120 minutes; HH100 group 29−35.1, mean 29.56 at 270 minutes) and recovered to approximately the pre-hypoxia temperature at 330 minutes in both groups. During the 2 hour post-hypoxia period, the Nest animals were 2−4 degrees cooler than the HH100 animals. Two-way repeated measures ANOVA showed significant effects of both time (p = 0.0014) and treatment group (p = 0.0049). Bonferroni post-hoc tests showed significant differences between treatment groups at 120 (p < 0.001), 180 (p < 0.01) and 210 minutes (p < 0.05). Two additional animals were monitored while exposed to room air in the warmed chamber for the two hour post-hypoxia period. Rectal temperatures of these animals were not significantly different from the HH100 animals (data not shown).

Figure 3.

Rectal temperature over time, means +/− standard errors. All pups exposed to 8% oxygen in 36°C chamber from 0−90 minutes. Nest group returned to mother at 90 minutes; HH100 group subsequently exposed to100% oxygen for 120 minutes in same chamber, then returned to mother at 210 minutes. Two-way repeated measures ANOVA showed significant effects of both time (p = 0.0014) and treatment group (p = 0.0049). Bonferroni post-hoc tests: significant differences between treatment groups (*) at 120 (p < 0.001), 180 (p < 0.01) and 210 minutes (p < 0.05).

Discussion

The results of this study indicate that long term outcome after exposure to 100% oxygen, as measured here by histological examination, was not different from that after exposure to room air in this model of perinatal hypoxic-ischemic brain injury. Our results further suggest that exposure to 40% oxygen after H/I may improve outcome, specifically in male animals.

As recently as 2000, international guidelines for resuscitation of newborn infants recommended the use of 100% oxygen (Kattwinkel et al., 1999). There has been increasing concern, however, that 100% oxygen may have adverse effects, particularly on tissues that may have already suffered oxidative stress (Lefkowitz, 2002; Saugstad, 2005). Two meta-analyses of the results of 5 published clinical trials of neonatal resuscitation with room air versus 100% oxygen have recently been reported (Davis et al., 2004; Saugstad et al., 2005). Although the inclusion criteria and statistical analyses are different between the two meta-analyses, both found a reduction in neonatal mortality with resuscitation using room air compared to 100% oxygen. The number needed to treat analysis predicted that “one death would be prevented for every 20 babies resuscitated with air rather than 100% oxygen” (Davis et al., 2004). Follow-up of one of these cohorts at age 18−24 months showed no differences in growth or neurologic dysfunction between children resuscitated at birth with either 21% or 100% oxygen (Saugstad et al., 2003).

There are several animal studies that support the use of room air rather than 100% oxygen for neonatal resuscitation. These studies have been conducted primarily in piglets, a large animal model selected for cardiovascular similarities to human newborns, as well as similarities in brain development. A variety of acute changes that may be related to oxidative injury have been reported. Goplerud (Goplerud et al., 1995) found persistent reduction of striatal Na⁺,K⁺-ATPase activity in piglets exposed to 100% oxygen for 2 hours after asphyxia, while those exposed to 21% oxygen returned to baseline activity. A transient increase in cerebral cortical extracellular hypoxanthine was found after hypoxemia followed by a 5 hour period of re-oxygenation with 100% oxygen, but not with 21% oxygen (Feet et al., 1997). Munkeby (Munkeby et al., 2004) reported increased extracellular glycerol and increased matrix metalloproteinase activity and expression in piglets resuscitated with 100% compared to 21% oxygen. Other investigators found no differences in reactive oxygen species, early growth response gene-1 (Tyree et al., 2006), or severity of brain injury after 4 days survival time (Rootwelt et al., 1992) between piglets resuscitated with 100% or 21% oxygen. In a short-term (4 hour) survival study, histological injury was greater in piglets resuscitated with 100% oxygen versus room air in the hippocampus and cerebellum, but the reverse was true in the basal ganglia (Domoki et al., 2006). No differences in neuropathological outcome were seen in neocortex (frontal and temporal) or pons.

With the exception of the localized and very short term histological outcome mentioned above (Domoki et al, 2006), in none of these studies were any outcome measures worse in animals or infants resuscitated with 21% oxygen, leading to the conclusion that the use of room air for neonatal resuscitation is as effective as 100% oxygen, and may be beneficial. The International Liaison Committee on Resuscitation concluded in 2005 that “there is currently insufficient evidence to specify the concentration of oxygen to be used at initiation of resuscitation” (ILCOR, 2006). The use of intermediate or variable concentrations of oxygen guided by pulse oximetry has been suggested as more appropriate (Hansmann, 2004; Milner, 1998), but there is currently little evidence to support this idea. A recent population-based study from Sweden compared the use of 40% versus 100% oxygen for neonatal resuscitation (Hellstrom-Westas et al., 2006). Swedish 1997 guidelines for neonatal resuscitation recommended the use of 40% oxygen for initial neonatal resuscitation, and while many hospitals followed this recommendation, a few chose to follow the International guidelines, which at that time recommended the use of 100% oxygen. Infants requiring resuscitation at hospitals that used 40% oxygen had higher 5 minute Apgar scores than those resuscitated with 100% oxygen; no significant differences were seen in 10 minute Apgar scores, neonatal death, seizures, or signs of hypoxic-ischemic encephalopathy. We are unaware of any animal studies prior to the current report that used concentrations of oxygen intermediate between 100% and room air.

In the animal model used in this study, while pups are exposed to systemic hypoxia after carotid ligation, the result is focal brain injury in the distribution of the middle cerebral artery (Rice et al., 1981). While acknowledging that cell death pathways and neuroprotective strategies may have different effects in focal versus global ischemic injury (for example (Kofler et al., 2006), it is encouraging that the lack of difference between post-hypoxia treatment with room air or 100% oxygen is similar to that reported in piglet models of global cerebral ischemia and in human infants.

Body and brain temperatures in rats at this age correlate with ambient temperature (Trescher, 1997), thus temperature must be monitored carefully. When one group of pups was returned to the mother at the end of hypoxia and the other remained in the warmed chamber, there was a 2 to almost 4 degree difference in the two groups during the two hours immediately post-hypoxia (Figure 3). The results for temperature differences in pups returned to mothers versus those kept in warmed chambers suggest that a small difference in body temperature in the post-hypoxia period may have a long term effect on brain injury. In the current study, post-hypoxia temperature was closely controlled, and no difference in degree of injury, as assayed here, was seen between 100% oxygen and room air groups. Many of the initial studies that showed a protective effect of post-ischemic hypothermia in neonatal animals looked only at relatively short survival times (one week or less) (Laptook et al., 1994; Thoresen et al., 1996; Thoresen et al., 1995; Towfighi et al., 1994; Young et al., 1983). There are few studies investigating the effects of temperature on long term morphologic outcome in similar 7 day old rat models. Studies reported in the literature are difficult to compare due to differences in what is considered normothermia, the degree of hypothermia, and length of time of exposure to hypothermia. Yager (Yager et al., 1993) found a protective effect at 30 days survival when animals were exposed to hypothermia (ambient temperature of 31°C or 34°C compared to 37°C) during hypoxia (3 hours), but no protection with 3 hours of post-hypoxia hypothermia. Trescher (Trescher et al., 1997) found that after a mild to moderate hypoxic insult, rectal temperature 3 hr after hypoxia (3 hours recovery at ambient temperatures of 32, 35 or 38°C, with rectal temperatures 34.2−38.6°C) correlated with the degree of brain injury at one week survival, but not at 4 weeks survival, suggesting that injury was delayed rather than prevented. Bona (Bona et al., 1998) reported that 6 hours of post-hypoxia hypothermia (rectal temperature of 32° versus 37°C for normothermia) reduced brain injury at both 1 and 6 weeks post-hypoxia. In the other direction, 6 hours of post-hypoxic hyperthermia (39°C compared to 36.5°C in the normothermic group) has been reported to increase brain injury at 30 days post-hypoxia (Fukuda et al., 2003). The effects of temperature on long term brain pathology appear to be influenced by multiple factors, including the severity of the initial injury, the degree of temperature difference, the timing of hypo- or hyperthermia relative to the hypoxic insult, and the duration of post-hypoxia hypo- or hyperthermia.

The current study also suggests that there may be sex differences in the response to oxygen treatment after perinatal H/I. It is now well accepted that adult female and male brains respond differently to ischemic injury (Hurn et al., 2005). While sex steroids certainly influence the response to injury, studies in cultured neurons and astrocytes also demonstrate sex differences in the molecular mechanisms of cell death (Du et al., 2004; Liu et al., 2007). Sex differences in perinatal brain injury have been less studied. While the initial studies using the neonatal rat model showed no differences in the degree of injury between males and females (Rice et al., 1981), most studies have used animals of either sex with insufficient power to analyze the data for sex differences. Recently, sex differences have been shown in the response to various interventions and perinatal H/I. Moderate hypothermia was found to have greater protective effect in females than males, 6 weeks after P7 H/I as evaluated by both neuropathology and sensorimotor function (Bona et al., 1998). Hagberg (Hagberg et al., 2004) reported that male, but not female, PARP knock-out mice had reduced brain injury compared to wild type animals 10 days after P7 H/I. Sex differences were also found in the response to administration of erythropoietin after middle cerebral artery occlusion in P7 rats (Wen et al., 2006). Erythropoietin treatment decreased infarct volume at 6 weeks post-injury and improved sensorimotor function to a greater extent in females than in males.

In conclusion, our results show no difference in long-term histological outcome as measured here between animals treated with 2 hours of 100% oxygen versus room air after hypoxic-ischemic injury in the 7 day old rat. This corroborates results from studies of post-hypoxia resuscitation of neonatal piglets and recent studies of neonatal resuscitation of human infants. In addition, our results suggest decreased injury in male animals that received 40% oxygen post-hypoxia. Given the current variability in neonatal resuscitation practices (Leone et al., 2006; O'Donnell et al., 2004) and the paucity of data on the use of intermediate levels of oxygen during resuscitation, further studies of the effects of different concentrations of oxygen on brain injury are warranted, and should have sufficient power to examine sex differences.

Abbreviations used

H/I

hypoxia/ischemia

H

hypoxia

HH

hypoxia followed by hyperoxia.