Serum neuron-specific enolase, magnetic resonance imaging, and electrophysiology for predicting neurodevelopmental outcomes of neonates with hypoxic-ischemic encephalopathy: a prospective study

Neonatal hypoxic-ischemic encephalopathy (HIE) is a type of encephalopathy caused by oxygen deprivation to the brain [1, 2]. It is mainly reported and studied in infants > 35 weeks gestation [1, 2]. Of note, the definition of HIE is more difficult in preterm infants, and preterm infants are not eligible for cooling; therefore, they have not been studied thoroughly. The reported incidence of neonatal HIE is 2–8 per 1000 live births in developed countries and 26 per 1000 live births in underdeveloped countries [2‐4]. It is most likely the result of an acute peripartum (most common) or chronic (during pregnancy) event [1, 2]. It can be caused by sentinel hypoxic or ischemic events such as severe placental abruption, umbilical cord prolapse, or shoulder dystocia, but most cases are idiopathic [1, 4]. Common neurologic sequelae of HIE include neurodevelopmental disabilities such as cerebral palsy, seizures, hearing loss, blindness, learning disabilities, and behavioral disabilities [1, 2, 5, 6]. Predicting the neurological prognosis would ensure optimal management and treatment and avoiding exposure to potentially toxic therapies [7, 8].

In the last decades, the Apgar score and Sarnat staging have been applied to predict prognosis [4], but follow-up studies demonstrated a low predictive value of Sarnat grading for the outcome of infants with HIE, especially moderate (Sarnat grade II) encephalopathy [9‐12]. The infant’s condition after therapeutic hypothermia, including neurophysiologic tests, amplitude-integrated electroencephalography (aEEG), biochemical markers, and neuroimaging, have been used to assess the severity of HIE and predict long-term outcomes [1, 2, 4, 8, 13, 14]. Indeed, the tests above are recommended by guidelines [1] and by the literature [2, 4]. In addition, cranial ultrasound, computed tomography (CT), magnetic resonance imaging (MRI) are valuable tools to determine the extent of ischemia and brain damage [8], with MRI being the most sensitive modality for determining the patterns of brain injury, without the use of radiations [13, 14]. Unfortunately, the results are influenced by the timing of the monitoring among studies [15‐17] or the different timing of the different examinations within a given study [15].

Neuron-specific enolase (NSE) is specific to central and peripheral neurons and neuroendocrine cells [18]. High blood levels of NSE are associated with poor outcomes after cardiac arrest, stroke, and pediatric traumatic brain injury [18]. Celtik et al. [19] showed that NSE could be used as a predictor of HIE severity. Hypothermia treatment for HIE might be associated with decreased NSE levels and possibly with neurodevelopmental outcomes [20], but the results need to be confirmed. Still, whether a combination of different methods could improve the prognostication of HIE remains to be examined. Some recent studies examined NSE levels, MRI, and aEEG on the outcomes of HIE within the same study, but without combining the different modalities [21, 22]. A study suggested that the combination of cerebrospinal fluid NSE, MRI, and EEG could predict the outcomes of HIE [23], but sampling cerebrospinal fluid in neonates is riskier than blood.

Therefore, this study aimed to examine the predictive value of serum NSE, aEEG, and MRI, alone and in combination, for the neurological outcomes in neonates with HIE.

Neonatal HIE is an important cause of mortality and morbidity [1, 2], but effective indicators for the early diagnosis of brain injury after asphyxia and prognosis are lacking [15‐17]. Therefore, this study aimed to examine the predictive value of serum NSE, aEEG, and MRI, alone and in combination, for the neurological outcomes in neonates with HIE. The results suggest that the combination of MRI, aEEG and 3-day NSE can predict the neurological prognosis of newborns with HIE. The combination of the three methods can improve the predictive ability for long-term neurobehavioral prognosis.

Previous studies examined the value of the Sarnat score that could be used to determine the prognosis of the infants; Polat et al. [42] and Liauw et al. [43] showed that the PPV and NPV of the Sarnat staging for the adverse outcome (disability and death) were 52 and 100%, while the PPV and NPV were respectively 45 and 0% in children with HIE grade II [43]. In the present study, the Sarnat staging had difficulty in assessing neurodevelopment at 18 months due to the multifactor dependency of HIE, as supported by the above studies. Therefore, more reliable methods are needed. The Thompson score has been shown to be of prognostic value in infants with HIE [44] but was not assessed in this study. In this study, HIE grading was based on Sarnat grading but could be affected by a variety of clinical factors. Sarnat grading is not good in predicting long-term outcomes, and this study discusses other predictive indicators. This is consistent with Shankaran et al. [45].

Elevated NSE levels have been reported after stroke, brain injury, cardiac surgery, cardiopulmonary arrest, and perinatal asphyxia [46‐48]. Kunda et al. [49] showed that important increases in NSE levels could be observed after seizures in patients with temporal lobe epilepsy, suggesting neuronal damage, but the cause-to-effect relationship could not be established. Lee et al. [50] suggested that NSE levels could be used for the differential diagnosis of seizures over syncope. In addition, Sheik et al. [51] showed that NSE is elevated in patients with acute diseases of the central nervous system, and that it was elevated in patients with seizures vs. those without. Therefore, NSE is associated with brain damage and seizures, but whether NSE levels increases before or after seizures remain to be determined.

HIE is a fetal or neonatal brain injury caused by decreased or suspended cerebral blood flow, with partial or complete asphyxia resulting from perinatal hypoxia. At the cellular level, decreased blood flow and oxygen delivery initiates a series of harmful biochemical reactions. Oxygen consumption interferes with oxidative phosphorylation, which leads to the conversion to anaerobic metabolism, rapid depletion of high-energy phosphate reserves, accumulation of lactic acid, and failure to maintain cellular function [1, 2, 4]. Cell membranes are damaged, and intracellular components of neurons, including NSE, are released into the cerebrospinal fluid and blood. In this study, increased NSE levels were observed at 3 days in neonates who eventually showed neurodevelopmental delay. It is supported by studies that suggested the value of NSE for HIE [52, 53]. Recent studies in asphyxiated neonates have correlated NSE with developmental outcomes, but the conclusions were conflicting because of different detection timing because it takes some time for the damaged cells to die and release their content [54], supporting the significant difference at 3 days but not at 1 day in the present study. Kecskes et al. [55] also reported that NSE levels at 3 days had the best predictive value. Still, it is unknown whether the higher NSE levels at 3 days compared with 1 day is the result of delayed NSE release in the blood compartment or the result of the ongoing injury process. This will have to be examined in future studies.

The abnormal electrical activity in brain regions can be diagnosed by aEEG. Nakamura et al. [56] reported that electrical changes could occur in damaged brain regions within 10–20 min after a hypoxic-ischemic attack, which can be detected by aEEG monitoring. The present study suggested that a persistent abnormal aEEG over 4 h and within 6 h after birth was associated with adverse neurodevelopmental outcomes. Still, the PPV of 6-h aEEG was poor (PPV of 42.9%), and good outcomes might occur despite an abnormal aEEG. Therefore, an abnormal aEEG within 6 h after birth might have no clinical value if it is used alone. Conversely, a normal 6-h aEEG has a good NPV (100%). These results are supported by a previous study [57], but Vasiljevic et al. [58] showed higher specificity and PPV than in the present study. Their study showed the pattern of abnormal aEEG correlated with the severity of HIE (P < 0.0001) and subsequent neurodevelopmental outcome (P < 0.001) [58]. They found that abnormal aEEG background patterns exhibited superior prediction of abnormal developmental outcomes at 12 months of age. Still, the discrepancies between the present and previous studies could be due to differences in monitoring time (within the first 72 h of life in their case [58]). aEEG is a useful tool for the monitoring and detection of seizures [59]. In neonates with seizures, more than two seizures detected by aEEG is predictive of progression to early-onset epileptic encephalopathy [60]. Still, the exact relationship between seizures and aEEG abnormalities could not be determined in the present study. Future studies will have to examine the exact role of seizures, duration of seizures, and seizure burden in infants with HIE.

Conventional MRI is an essential tool for assessing the neonatal brain injury and degree of injury, but sometimes T1- and T2-weighted images can appear normal during the few days after birth or show only subtle findings that might be difficult to interpret or to distinguish from normal maturational phenomena in the neonatal brain. The most appropriate time for examination is 4–7 days after birth [35, 36]. The present study showed that in term infants with HIE, conventional MRI detection 4–7 days after birth can be used for the relatively accurate prediction of neurodevelopment outcomes at 18 months old (sensitivity, specificity, PPV, and NPV, respectively were 94.44, 71.88, 65.38, and 95.83%).

Still, whether a combination of NSE, MRI, and EEG could improve the prognostication of HIE remains to be examined. Indeed, Leon-Lozano et al. [21] and Shim et al. [22] recently reported NSE levels, MRI, and aEEG (all assessed within the same study) on the outcomes of HIE, but without combining the different modalities. A previous study published by Ezgu et al. in 2002 [23] suggested that the combination of NSE, MRI, and EEG could be used to determine the prognosis of HIE, but they measured NSE in the cerebrospinal fluid, which is less practical than from the blood. In the present study, consistent relationships between the outcomes at 18 months and the injury degree on conventional MRI, aEEG results, and serum NSE levels of 3 days after birth (r_s = 0.719, P < 0.001) were observed. When dichotomizing the NSE levels based on the best cut-off value and when considering MRI and aEEG as normal/abnormal, combining the three test results in better sensitivity, specificity, PPV, and NPV than for each test alone. This possible predictive tool should be validated in a larger group of patients.

In this study, 27% of the children with mild HIE had a severe delay. Most previous studies used the traditional clinical Sarnat grading system, in which infants with mild Sarnat grading had a good prognosis and no long-term disability. For this reason, many studies did not carry out a systematic examination and long-term follow-up on neonates with mild HIE. Still, more and more studies are finding that they can experience major disability, and when followed up to 18 months, about 25% of infants with mild HIE have abnormal outcomes, defined as death or movement and/or developmental delay [61, 62]. Nevertheless, it is not clear why a minor injury would lead to such severe adverse outcomes. It might be that early hypoxia injury will change, worsen, and expand over time. Since a significant proportion of children with mild encephalopathy have been reported to have significant long-term neurological sequelae, more measures are needed to evaluate brain injury and better predict prognosis in addition to Sarnat grading.

There are limitations of this study. The sample size was small and from a single center. Long-term outcomes were not followed. A larger sample size, multiple centers, and longer clinical follow-up are needed to confirm the findings. In addition, additional indicators of electrophysiological and biochemical changes of neurons after hypoxia and ischemia should be explored. Of note, the MRI examinations were read by a general radiologist, not a pediatric radiologist or a neuroradiologist. Due to copyright issues, Bayley III cannot be translated, standardized, and validated in Chinese for the time being; therefore, Bayley II was used. Finally, in this study,the mortality among moderate/severe HIE infants was much lower than seen in the RCTs of therapeutic hypothermia and in subsequent prospective studies, part of the reason is that the special and complex doctor-patient relationship in China. When there is subjective evaluation for disease severity, Chinese doctors tend to classify them as severe cases, which is relatively common and may be different from other countries, may have had some effect on the results.

In conclusion, in the earlier stages after birth, aEEG, MRI, and 3-day NSE levels can be used to predict the 18-month neurodevelopmental outcomes of children with HIE not treated with hypothermia treatment. These tests might have a use in predicting prognosis in full-term infants with HIE, providing them with more accurate management strategies.