Cerebral rScO2 Measured by Near-Infrared Spectroscopy (NIRS) During Therapeutic Hypothermia in Neonates with Hypoxic-Ischemic Encephalopathy: A Systematic Review

Perinatal asphyxia is one of the main causes of neonatal mortality and medium- and long-term neurological sequelae in the paediatric age group [1]. For low- and middle-income countries and clinical settings, it is defined as failure to initiate or maintain breathing at birth [2]. The current treatment for neonates with moderate to severe asphyxia is therapeutic hypothermia initiated in the first six hours of life [3] to reduce mortality, morbidity, and the severity of neurological sequelae [4]. However, medium- and long-term sequelae remain high [5].

Therefore, the early detection of pathophysiological alterations at the neurological level during hypoxic-ischaemic encephalopathy before and during treatment that can predict brain damage would allow therapeutic interventions to be instituted in the early stages of the disease, establish follow-up before discharge from the unit in high-risk neurological programmes, and deepen the knowledge of the mechanisms of neurological injury in the phases of asphyxia. In this area, neuromonitoring during the 72 hours of cooling and rewarming has been established as an option to recognise cerebral hypoperfusion, encephalopathy, abnormal electrical activity, and, in general, physiological, clinical, and/or biochemical variables that serve as biomarkers for relevant identification [6].

Currently, neuromonitoring is centred on electroencephalography (EEG), amplitude-integrated electroencephalography (aEEG), electroencephalogram, and video telemetry [7]. These electrophysiological tests during cooling have shown usefulness in defining prognosis and the presence of brain injury in the first six hours of treatment [6]. However, abnormal tracing after six hours, especially in the first 24 to 36 hours, has a low predictive value [8]. On the other hand, the presence of abnormalities in brain magnetic resonance imaging (MRI) in the first week of life has a high predictive value for neurodevelopmental alteration and neurological sequelae in the medium and long term. However, its findings are late and changeable in the evolution of perinatal asphyxia, so when the alteration is found in the images, there is already an established neurological injury [9].

Near-infrared spectroscopy (NIRS) is a non-invasive, portable, and reproducible bedside monitoring technique within the neonatal intensive care unit without potential harm to the neonate [14]. It provides continuous, early, and timely information on regional cerebral haemoglobin oxygen saturation, cerebral blood volume, and cerebral oxygen supply/metabolism ratio, making it a potential technique for neuromonitoring in neonates, especially in brain tissue hypoxia and associated injuries [15]. Continuous monitoring of cerebral regional oxygen saturation (rScO2) would allow timely diagnosis and may have the ability to predict early brain injury during the early stages of neonatal asphyxia. In addition, together with other clinical monitoring variables, it could identify pathophysiological alterations in cerebral blood flow [9]. Studies in animal models show the potential role of NIRS in the timely detection of brain injury [9]. Likewise, research in neonates is promising for the association between changes in and values of rScO2 during hypothermia and the future presence of neurological injury [10].

Integrating different tests when monitoring brain functions, from the electrophysiological perspective, metabolism, and oxygenation, would allow the optimal and timely identification of the neonate at risk of injuries and neurological alterations [9]. Additionally, predicting neurological outcomes among neonates with hypoxic-ischaemic encephalopathy (HIE) has become a critical and ongoing challenge. This study aimed to conduct a systematic literature review of published data on rScO2 monitoring, measured by NIRS, during hypothermia treatment in neonates with perinatal asphyxia to predict short- and long-term neurological injury. The secondary objectives were to characterise the quality of the studies entered and to explore heterogeneity. This study helps to characterise the current status of rScO2 by NIRS in neurological injury in neonates with asphyxia and may provide a basis for future research design.

The study protocol was registered in PROSPERO (International Prospective Register of Systematic Reviews) with CRD code 42023395438.

n = 380 articles were collected from databases in the initial search, of which 181 were eliminated as duplicates when entered into the web tool Rayyan®, resulting in a total of 199 articles being discriminated by title and abstract. From this process, 27 articles met the inclusion criteria for full-text reading, and finally, 15 articles were selected for extraction and analysis of the information. Among the reasons for discarding the remaining ones were incorrect results (n = 5), full text not found (n = 1), types of publication and/or different types of study (n = 5), and incorrect population (n = 1). See Figure 1.

Figure 1.

The methodology of the studies was 100% analytical observational cohort studies, 67% prospective, and 33% retrospective. All met the sampling power target set by the authors. The total number of neonates studied was n = 362, including term and near-term neonates. In general, NIRS measurements were performed in frontal, parietal, or frontoparietal locations, of which seven were bilateral and eight unilateral, three parietal, four frontoparietal, and eight frontal, the latter being the most common location.

The neurological outcomes studied included short-term (n = 6) and long-term (n = 5), and n = 4 included both short-term and long-term outcomes. For the identification of short-term outcomes, brain MRI was used in the first days of life, mostly in the first week, in nine studies; amplitude-integrated electroencephalogram (aEEG) in two studies, and only one study evaluated short-term neurodevelopment using the Thompson scale [13]. On the other hand, for the identification of long-term outcomes, neurodevelopment was assessed using BSID III scales (Bailey Scales of Infant Development Version 3) in four studies and GMDS (Griffiths Mental Development Scales) in four studies, while one study implemented Capute and Mullen scales (Table 1).

Overall, the quality of the studies was adequate, with an average of 78% of the STROBE list for cohort studies (Figure 2). Twenty-two percent of the remaining articles are at high risk of bias, mainly due to the omission of study biases and the failure to describe sample sizes.

Figure 2.

The systematic review addressed the use of neuromonitoring of rScO2 by NIRS during hypothermia treatment in the prediction of and relationship to short- and long-term neurological injury in infants with perinatal asphyxia. The synthesis of this information shows that an increase in cerebral blood flow between 24 and 48 hours is related to an increase in the probability of short- and long-term neurological lesions. Likewise, although the risk of bias in the included observational studies is low, there is high heterogeneity in the information, especially in the definition of outcomes, methodologies, and measurement times.

The increase in cerebral blood flow, subrogated to the elevated value of rScO2 measurement, is explained from the point of view of pathophysiology. During neonatal asphyxia, changes at the cellular level occur at different stages of asphyxial injury. In general, the initial chain of events is composed of hypercapnia, hypoxemia, and acidosis that generate a loss in cerebral autoregulation and an increase in initial cerebral blood flow; simultaneously, cellular changes occur in the CNS such as necrosis, secondary reperfusion, and finally apoptosis [26].

After the initial insult (asphyxia), a primary energy failure is generated, accompanied by the consumption of high-energy phosphates and glucose, with failure of the Na/K ATPase pump and mitochondrial function, favouring oedema and early cell death (necrosis); this initial alteration generates different alterations in factors related to maintaining the self-regulation of brain metabolism, among which is an increase of hydrogen ions as a consequence of an anaerobic cycle of glucose consumption, which in the absence of oxygen enters the Cori cycle generating lactic acid and releasing hydrogen ions to the extracellular space. When these are captured by the cell, an exchange of ions of equal charge takes place, producing hyperkalaemia, leading to an increase in extracellular potassium that decreases the repetition of smooth muscle cell action potentials, thus keeping the cell hyperpolarised and generating vasodilatation. In addition, phosphorylation of ATP to ADP secondary to pump failure generates adenosine, which binds to P2Y receptors and generates equal vasodilation. Similarly, hypercapnia and increased carbon dioxide pressure play an important role in vasodilation [26].

All this results in a phase of hyperaemia and increased cerebral blood flow, a situation that explains the increase in the NIRS value found in the different studies correlating this pathophysiological characteristic. Subsequently, a second energetic failure occurs in which excitatory amino acids are released with increased oxidative stress due to calcium influx, which perpetuates the production of proteases and the release of free radicals, favouring the induction of pro-apoptotic factors that perpetuate late cell death [13].

Therefore, considering the factors involved in the pathophysiological process, their correlation with the initial elevation of rScO2 during treatment, and together with the measurement of clinical biomarkers such as severe acidosis, hypotension (accompanied by an immature vascular autoregulation system in the neonate), hypoxemia, hypercapnia, and electrolyte disturbances such as hyperkalaemia, these clinical variables and their correlation with cerebral blood flow (rScO2 by NIRS) and the state of cerebral autoregulation during the disease are important so that the clinician can perform interventions at the patient’s bedside to maintain these physiological variables in adequate ranges and protect cerebral blood flow in normal ranges. This would help to maintain an acid-base and hydro-electrolyte balance, understanding that these are influential elements in cell damage [26].

Similarly, it is important to understand the role of brain metabolism during neonatal asphyxia, and this is reflected in other NIRS measurement indices, such as the change in cytochrome c oxidase oxidation as an indicator of mitochondrial metabolism and ATP synthesis. Bale G. et al. (2019) report a significant relationship between decreased oxygenation and decreased cytochrome C oxidase oxidation in neonates with severe neurological injury, indicating the coexistence of a mismatch between oxygenation and metabolism at the cellular level in EHI with an unfavourable outcome [26]. This is explained in the chain of events during secondary energy failure, in which increased oxidative stress generates a permeability pore in the mitochondria, which, in addition, reaches a fully oxygen-dependent redox state, leading to the release of cytochrome C and, consequently, apoptosis-inducing factors [13]. Therefore, asphyxiated neonates that reach secondary energy failure and the apoptotic process will present worse neurological outcomes. Therein lies the importance of early interventions to avoid this stage, as well as the identification of the nature and severity of the insult, considering that this could be antepartum and, therefore, produce greater progress in the damage.

Likewise, it is essential to maintain optimal mean arterial pressure; small changes in arterial pressure will affect the cerebral blood flow autoregulation curve in scenarios of loss of cerebral autoregulation [26].

Authors such as Lee J. et al. (2017) confirmed this in their study [27], showing that NIRS values increased simultaneously with altered mean arterial pressure measurement. To improve neurological outcomes, both hypotension and hypertension states should be avoided in this group of patients. The clinician should constantly correlate the blood pressure monitoring data with the rScO2 value during treatment and make frequent adjustments to avoid damage to the neonate’s cerebral autoregulation [28].

Authors such as Marin T. et al. (2011) [28] have proposed the utility of NIRS in neonates for the continuous measurement of regional tissue oxygenation in different organs, mainly the brain, kidney, and mesentery, which reflects the perfusion status and allows physicians to directly monitor fluctuations in this perfusion in real-time. This statement is corroborated by the studies reviewed in this investigation (see Table 1) since they agreed on the usefulness of NIRS as a marker of cerebral blood flow (CBF), which indirectly gives us a measure of cerebral tissue perfusion.

Studies evaluate the association of brain NIRS with short-term and long-term outcomes. In the short term, in the investigations included in the systematic review, neurological injury is defined especially by brain MRI findings before discharge, usually within the first week of life. MRI is currently a widely recommended tool with standardised data to determine the pattern and severity of brain damage and the prognosis of the neonate with HIE [29,30]

The characterisation of findings related to neonatal asphyxia and its severity spectrum have been defined in some common injury patterns, such as deep grey matter lesions, posterior internal capsule lesions, cortical infarction, white matter hyperintensity in T2, and punctate lesions in white matter [31].

In most of the included studies, neuroimaging is performed after hypothermia, during the fourth or fifth day of life, and/or during the second week of life, as in the case of Wisnowski’s study [32]. However, it is important to highlight that the study by Liu W. et al. [33] reports that the specificity and sensitivity of MRI are better within two weeks of birth. Given the changing and late-onset pattern of MRI, the usefulness of NIRS measurement as an early predictor of brain injury is raised in the literature included in the systematic review. Thus, for example, the research by Szakmar E. et al. [22] shows that neonates with grey matter injuries in MRI correlate with higher rScO2 values on the second day of hypothermia, and in the study by Niezen C. et al. [14], the NIRS value of > 90% at 48 hours >90% is associated with severe brain injury by MRI. Thus, the measurement of NIRS before, during, and after therapeutic hypothermia can be proposed as an early predictive biomarker of neurological outcomes compared to MRI. This considers that according to the findings of this study, high NIRS values during the first 24 to 48 hours are related to unfavourable neurological outcomes, while brain MRI during the first 24 hours of life has limitations when it comes to fully identifying the extent of the lesion [34].

On the other hand, other authors also describe short-term outcomes using electroencephalograms (EEG). Continuous use of EEG in neonates with moderate to severe HIE helps to estimate the bioelectrical function of the brain in addition to detecting subclinical seizures [35]. In the study by Del Rio et al. [36], EEG can predict brain injury during the first 72 hours of life. In contrast, Niezen C. et al. [14] found that NIRS is a better predictor of neuronal injury at 72 hours than EEG. Since EEG can show abnormal electrical patterns in the brain, it is a useful tool at the bedside for early intervention, if necessary. However, the combination of bedside neuromonitoring tools within the neonatal unit, allowing a window into different physiological variables in the course of the disease, will be better than the isolated use of each in terms of neurological prediction and decision-making [8, 35].

Concerning long-term outcomes, the included studies assess neurodevelopment in the first 36 months of life using different instruments, all validated for this purpose. They specifically use the BSID III scales (Bailey Scales of Infant Development Version 3) between 18 and 36 months and the GMDS scale from three months to five years (Griffith Mental Development Scales). Regarding the BSID III scale, a cut-off point of < 85 has been described to define neurodevelopmental delay, and a close relationship has been found between low scores on the BSID III scale in cognitive, language, and motor domains in children with moderate hypoxic-ischaemic encephalopathy [37]. According to the data extracted in this systematic review, there is evidence of a relationship between high brain NIRS values and low scores in neurodevelopmental scales, which would make brain NIRS monitoring in the first hours of treatment and altered values useful for predicting neurodevelopmental alterations between 18 and 36 months. In their studies, Oliveira C. et al. demonstrated NIRS values at 48 hours above 83% correlate with a lower score in the BSID III and GMSD scales [16], and Toet et al. described a trend of increased cerebral rScO2 and decreased FTOE in the first 24 hours for neonates with neurodevelopmental disorders at two years of life. However, other studies have reported the limited ability of such scales to identify more subtle deficits in the cognitive spectrum, so children should be evaluated over the longer term, including school performance, to establish the impact of EHI on neurodevelopment across the lifespan [38]. This suggests that alterations in brain NIRS values may be related to neurodevelopmental delay, but normal NIRS values and scores on neurodevelopmental scales do not rule out the coexistence of some degree of compromise. Given the limited information found in other studies about NIRS values and long-term neurodevelopment, further research on this topic is suggested.

Among the strengths of this systematic review are the quality of the studies and the low risk of bias. The data obtained are reliable and extrapolatable, which generates good inferences from these data with a low risk of bias. However, one limitation is that most of the studies did not propose a sample size for the proposed outcome. Another limitation was the heterogeneity in different variables that would allow standardisation of the data and results. In this sense, we found significant variability between studies in the value of NIRS as a predictor of acute brain injury, the time during treatment that can be a predictor of the value obtained, and methodological aspects of the measurement such as the site of placement for the measurement and the instrument used. Therefore, working on the standardisation of the measurement is recommended to improve the interpretation of the result and its impact on patient care. In addition, heterogeneity was also found in the definition of short- and long-term outcomes. As brain MRI and neurodevelopmental scoring scales are not standardised across studies, it is necessary to define a standardised method for measuring brain injury in MRI as well as a scale for measuring neurodevelopment in asphyxiated neonates.

A relationship is observed between high values of rScO2 in cerebral NIRS during hypothermia treatment, especially in the first hours of treatment, and the prediction of short-term (brain MRI) and long-term (neurodevelopmental alterations) neurological injury. Neuromonitoring with NIRS as a biomarker is a useful bedside tool to predict brain injury in neonates with moderate to severe HIE. It also provides real-time information for timely and early follow-up and therapeutic interventions in the neonatal care unit.