Commentary to the in-focus issue “Perinatal brain injury leading to later neurodevelopmental disorders: Early detection and treatment options”

This issue introduces the reader to a range of important issues that involve neonatal encephalopathy: from different animal models to neuroprotection and clinical biomarkers. The collection of papers is important because the reader obtains a bird’s eye view of the landscape of perinatal brain injury, and also, the various nuances in dealing with translational and clinical issues.

Robinson and colleagues report on the beneficial effects of postnatal intraperitoneal erythropoietin administration on adult cognitive deficits in a rat model of placental insufficiency (Robinson et al., 2021). This article is notable not only for focusing on functional outcomes, but also for focusing on adult cognitive outcomes after a prenatal insult. Furthermore, the proper use of touch screen technique to test the animal behavior and cognitive function is presented. The translational implication of adult outcomes suggests that the 5-year cycles of NIH funding may make it difficult to establish clear-cut benefits of a promising postnatal neuroprotectant following a prenatal insult. The reader is left to wonder whether the absence of benefit in death or severe neurodevelopmental impairment at 2 years of age in the EPO trial in 24- to 32-week premature infants (Juul et al., 2020) and in the recently published clinical trial of high-dose-erythropoietin-for-asphyxia-and-encephalopathy (HEAL) trial given intravenously (Wu et al., 2022) could be because a longer follow-up period was not possible. Importantly, Robinson’s article uncovers sex differences in adult cognitive outcomes and the deficits in the visual system. To estimate the severity of the 60-min uterine ischemia at E18, the previous publication by the group (Jantzie et al., 2013) mentioned a fetal loss of 23% but did not mention any newborn deaths. A severe prenatal insult would usually result in postnatal deaths. The implication of possibly all newborn survivors reaching adulthood in the present study is that the status of brain injury probably updated to a mild-to-moderate severity at the time of postnatal EPO administration. Also, the dosage used in the Robinson’s article, 2000 U/kg dose i.p. for 5 consecutive days in P1–P5 rats, is roughly double when compared to that used in the HEAL trial in human newborns, 1000 U/kg i.v. dose on 1, 2, 3, 4, and 7 days (Wu et al., 2022). Clearance of EPO is approximately three times faster in rats and equivalent dosages in humans should be one-third than that of the dose given to rats (Nguyen et al., 2018; Woo & Jusko, 2007). The reader is left to wonder whether the dose used in humans was too high as neuroprotection is generally shown at the dosages mentioned in Robinson’s article. The possibility that EPO has a bimodal distribution in neuroprotection should be explored in future as it is possible that a toxic high-dose effect could negate the beneficial low-dose effect in humans. There is some evidence to support this view. There was actually a significant, albeit very small, increase in mean number of serious adverse events and percentage of children with at least one adverse event with EPO in newborns in the HEAL trial (Wu et al., 2022), raising the possibility that the dosage used in humans may have been on the high side.

Dettman and Dizon present an important review that has implications for all those who work in hypoxia-ischemia in newborn brain in humans as well as in animal models (Dettman & Dizon, 2021). They trace the origin of the association between preterm chronic lung injury and white matter impairment to a central role of oxygen levels. Brain injury may depend not only on tissue pO₂ levels but also on the vulnerability of the developmental stage of the cells to the fraction of inspired oxygen (FiO₂). This article raises more questions on etiological role of tissue oxygen in white matter injury, and points to the need to do more fundamental research on the tissue pO₂ levels in white and gray matter in the developing brain. After all, human brain tissue pO₂ may be higher than that of murine brain (Nortje & Gupta, 2006). The review provides a succinct summary of findings on models of hyperoxia that use 75%–80% FiO₂ in cell culture and in vivo animal models. One is unsure if there is a linear continuum from 21% to 80% in tissue pO₂ for cellular injury and cell death. The choice of 75%–80% FiO₂ is somewhat arbitrary as there is little evidence to suggest (a) there is a linear progression of cellular injury and death with increasing FiO₂ greater than 21% (room air), or (b) at least 75% FiO₂ is needed to trigger cellular injury or death. Dettman and Dizon have previously shown white matter injury using a single hit of hyperoxia or a double hit of hyperoxia and intrauterine growth retardation (Chang et al., 2018). Like other translational laboratories, the authors had to choose a level that causes enough damage to cells in an animal model. This is one of the known limitations of the hyperoxia models that must use a certain threshold of injury. There is probable need for more appropriate models mimicking the bronchopulmonary dysplasia states, which are more likely to have alternating hypoxia and hyperoxia periods of varying intensities. There is a series of questions that still need to be answered. What is the threshold of injury in human brain injury? Is the threshold of injury triggered by 75%, 80%, or some other FiO₂? Is the area under the curve of hyperoxia-with-time exposure important in brain injury? Is there need for repetitive bursts of hyperoxia to trigger brain injury? Is there need for hypoxic bursts in addition to hyperoxic bursts?

Shi et al. introduced a more translational rabbit model mimicking the transition from partial to total acute placental insufficiency (“Partial+Full”) (Shi, Luo, Jani, et al., 2021) and compared it to sudden onset of total placental insufficiency (“Full”) that the Tan laboratory has used before (Derrick et al., 2004; Drobyshevsky et al., 2007). It is unknown in human placental abruption how the placental insufficiency progresses with the progress of the separation of the placenta in the evolution of partial to total placental abruption. There are four important findings of this study. First, in this extensive study at two different ages, 70% and 79% gestation, they show no differences in death and motor deficits between the Partial+Full and Full models. However, the suck-and-swallow test was the only significantly different individual neurobehavioral test, with the Full model being worse, and after 24 h of insult, the brain cell viability was significantly less in Partial+Full. Second, the main advantage of the Partial+Full model is that neuroprotectants given to the mother after the onset of fetal bradycardia can still cross the placental barrier and reach the fetus. Third, this study took years to accumulate the numbers to perform a Phase III clinical trial design with testing for a difference of 15% between the two models and power of 81%. Very few laboratories use Phase III clinical trial design with power of 80% when using in vivo animals. But it is imperative sometimes to find mechanisms and pathways by conducting a large-number animal trial. This may be the biggest contribution of this study. In this study, the actual power was 58% with the distribution in the primary outcome of death and motor deficits. In individual tests, some tests reached 97% power, given the differences in variance. In a previous study by the same laboratory (Vasquez-Vivar et al., 2020), the importance of performing a Phase III clinical design was also emphasized, by showing a four-way interaction between tetrahydrobiopterin levels in four brain regions and hypertonia, which would not have been found with a less-powered study. Third, this was the first study to comprehensively investigate correlations between two individual sensory/motor/dystonia tests. By enrolling enough numbers of animals (1004 kits) to do a 80% power study, it becomes very clear from looking at the areas covered in orange and red in Tables 3 and 4, that there is a weaker correlation between hind limb tone and locomotor variables in Partial+Full at both ages, but a stronger correlation between odor response and tone at E25 in Partial+Full. These findings are important because they suggest that the injury may affect different regions of the brain in the two models. In future, a coupled MRI study would elucidate differences in the brain connectome but finding funding for a large MRI animal study continues to be a huge problem.

The biggest take away for the reader in the fourth article by Shi et al. is the concept of not only looking for statistical significance in human studies, but also replacing it with a focus on biological significance (Shi, Luo, Deol, et al., 2021). This estimation approach for biological significance has more clinical usefulness than traditional linear statistics. Shi et al. have tried to help clinical decision-making as to the usefulness of noninflammatory biomarkers by introducing the concepts of “strong” and “doubtful” biomarkers in a review of cerebrospinal fluid (CSF) biomarkers in newborn humans. The reader also can estimate the strength of the biomarker, which greatly helps the clinician in making prognostic statements. In order of decreasing strength, the strong biomarkers are creatine kinase, xanthine oxidase, vascular endothelial growth factor, neuron specific enolase, superoxide dismutase, and malondialdehyde. Intuitively, CSF is much better source of biomarkers than blood. The finding of some promising biomarkers is tempered by two factors. One factor is that most of the techniques reviewed are somewhat dated. With newer techniques that are presently available, the oxidative stress markers would be even more precise and informative, especially for xanthine oxidase, superoxide dismutase, and better biomarkers than malondialdehyde. The other factor is that unfortunately, the practice of performing spinal taps in neonatal encephalopathy is becoming more and more rare. Perhaps this is partly because of the unfortunate characterization of most cases of neonatal encephalopathy associated with signs of perinatal asphyxia as “hypoxic-ischemic” encephalopathy, ascribing hypoxia-ischemia as the etiological cause when there is really no empiric proof that such an etiology existed. It may be that to probe the etiology, the practice of a routine spinal tap may be necessary in most cases, especially when there is an absence of “sentinel” events prior to delivery (Okereafor et al., 2008).

Hopefully, this issue has met its purpose of providing some insight to the readers while stimulating the generation of more questions than answers.