Metabolomics and perinatal asphyxia
Ernesto d’Aloja, Emanuela Locci, Antonio Noto, Matteo Nioi, Giovanni Bazzano, and Vassilios Fanos
The incidence of neonatal encephalopathy worldwide is still elevated, with a magnitude of roughly 3 per 1,000 live births (1).
Neonatal encephalopathy (2–4) and hypoxic-ischemic encephalopathy (HIE) (5) are frequently used interchangeably, even though they may sometimes refer to different nosographic entities. This in-depth, and sometimes rough, debate about the terminology relies on the clinician’s ability to identify the causative event of this complex neurological syndrome. A major concern are the potential impediments that an inappropriate label may cause to a truly collaborative research in the field of preventive and therapeutic interventions. Moreover, the use of the HIE label may sometimes lead, in the malpractice scenario, to a fallacious judicial understanding concerning the cause of the newborn damage. If so, midwives, physicians, and hospital convictions may rely on the hypothesis of a hypoxic event during labor or in the perinatal period (during the first 6 “gold” hours). For this reason, and only for the purpose of this chapter, the “omni-comprehensive” term of perinatal asphyxia (PA) has been adopted, although if we are well aware of this label limitation.
In this scenario, metabolomics seems to be a very promising and helpful tool to unravel this polymorphic syndrome (6). Either the metabolomics profile or specific metabolites expressed in one or more biofluids have been shown to be positively related to PA.
Metabolomics relies on the detection, identification, and quantification of the low molecular weight metabolites present in a biofluid or tissue and on the study of the modifications induced by any physiopathological stimulus. It represents a multiparametric approach that gives a holistic view of the organism’s metabolic status. Considering the complex and multifactorial nature of PA, the use of metabolomics that allows for the screening of a wide variety of molecules from different biological pathways seems more promising than the use of single biomarkers, which may be neither adaptable nor adequate to appropriately identify the damage and to predict the outcome. The identification of specific metabolomics profiles may indeed help to improve early diagnosis, improve classification, and provide suitable treatments and follow-up. Different biofluids have been investigated in both animal models and human studies of PA by means of nuclear magnetic resonance (NMR) spectroscopy and mass spectrometry (MS) coupled with gas or liquid chromatography (GC or LC). The information gained by both analytical platforms is complementary in order to achieve the identification of the largest number of potential biomarkers or the most representative profile associated to the PA condition. In the last two decades, approximately 30 scientific papers appeared in the literature dealing with this issue. The main objectives of these studies were the assessment of the PA severity, the response to the reoxygenation with different protocols, the biological response to therapeutic hypothermia (TH) treatment, and the investigation of “potential” biomarkers of outcome. The main results of these studies are detailed in Table 31.1 (animal models) and Table 31.2 (human studies) and summarized in the following paragraphs.
Several models, from the simplest murine to the more complex swine or nonhuman primate, were investigated. Different biofluids, such as urine, plasma, cord blood, and cerebrospinal fluid (CSF) and tissues, along with brain, retina, and choroid were analyzed by both NMR and GC- or LC-MS.
Hypoxic-ischemic brain injury and the response to TH in neonatal murine models were investigated focusing on the NMR analysis of brain tissues (7–10). Studies were conducted ex vivo on superfused rat brain slices, where asphyxia was induced by oxygen-glucose deprivation, and on whole brain extracts of mice that underwent right carotid occlusion. In both models, several hypoxia-related metabolomics alterations, mainly associated to changes in tricarboxylic acid (TCA) cycle fluxes and neuron-glial differences, were identified, and the effect of different hypothermic protocols was analyzed.
The effects of mild and severe hypoxia were investigated by 1H-NMR in fetal lamb CSF (11). Fetal hypoxia was achieved by decreasing the maternal FiO2, and CSF was withdrawn from the fetal cisterna magna in utero. During severe hypoxia (pH < 7.1), metabolites of cerebral energy degradation and of neuronal cell damage were altered, while the metabolomics effect of a prolonged (2 hours) mild hypoxia (7.23 < pH < 7.27) was characterized by extensive cerebral energy degradation.
The effect of asphyxia and resuscitation was widely studied on newborn piglet urine and plasma samples using multiple analytical platforms. In a population of 33 asphyxiated piglets analyzed via targeted LC-MS, the plasmatic ratios of alanine and glycine to branched chain amino acids (BCAAs) were highly correlated with hypoxia duration (12). Several reoxygenation protocols were used, i.e., 100% oxygen, 100% followed by room air (RA, 21% oxygen), or simply RA. While lactate was reduced in a similar extent independently of the resuscitation protocol, TCA cycle intermediates and acylcarnitines, accumulated during hypoxia, were reduced at a different rate, 100% oxygen being associated to a slower cellular metabolic recovery. An untargeted LC-MS approach was applied on the same asphyxiated piglet model (13). Immediately after the hypoxic insult, plasma choline, fatty acids, hypoxanthine, and other intermediates of purine and pyrimidine metabolism were found increased, while approximately 2 hours after RA reoxygenation, the metabolomics fingerprint was almost completely recovered. A predictive metabolite score involving three plasma metabolites (choline, 6,8-dihyroxypurine, and hypoxanthine) was shown to have maximum correlation with hypoxia time (14). Three precursors of cytidine-di-phosphate-choline (CDP-choline), originating from phosphocholine, were proposed for grading the intensity and the duration of tissue hypoxia in plasma samples (15). A significant increase in CDP-choline was also observed in piglets’ retina after an intense period of hypoxia (16). The simultaneous study of piglets’ retina and choroid indicated that the two ocular tissues are characterized by a different metabolomics response to hypoxia (17).
Urinary 1H-NMR metabolomics allowed the identification of several metabolites able to differentiate between hypoxic and normoxic piglets (18–21). They were mainly involved in cellular energy metabolism and amino acid deregulation following hypoxia. Different reoxygenation protocols (18%, 21%, 40%, and 100% oxygen) led to a distinct group-related metabolomics profile, being the RA one associated with a higher survival rate and a shorter resuscitation time.
Urine and plasma samples of 125 asphyxiated piglets resuscitated with different cardiopulmonary resuscitation protocols were examined (22). The data obtained from the two matrices showed only poor correlation. The authors concluded that urine is not the ideal biofluid for real-time monitoring of acute conditions such as hypoxia, being urinary metabolomics modifications delayed and dependent on renal function and individual clearance patterns of the metabolites.
PA was also modeled in nonhuman primates (23–26). Moderate-to-severe PA was induced in utero by umbilical cord occlusion prolonged for two different periods (15 and 18 minutes). GC-MS was used to analyze blood from the fetuses (at baseline by umbilical cord venipuncture) and from newborns (by aortic catheter placed in situ before delivery through umbilical artery) at predetermined time points (5 minutes, 24, 48, and 72 hours after birth) encompassing delivery, resuscitation, and TH. Authors identified potential biomarkers characteristic of each treatment group, and they correlated the metabolomics alterations with the severity of brain injury (biomarker of damage) and with death or cerebral palsy (biomarker of outcome).