A COMPREHENSIVE PROGRAM OF OPTIMIZATION THE BRAIN OXYGEN STATUS IN NEWBORNS WITH RESPIRATORY DISORDERS BASED ON TARGETING REGIONAL OXYGENATION

NIRS  provides  the  ability  to  obtain  information  about  the  brain  oxygen  status  non-invasively.  Therefore, potentially, this technology can make it possible to correct therapy that affects the blood and oxygen supply to the brain. The aim of our study was to develop a guideline for a dynamic monitoring of the oxygen status based on regional oxygenation data, as well as an effective therapeutic tool for newborns with respiratory distress. 78 newborns with  respiratory disorders were  included  (37  late preterm and 41  full-term)  in  this observational  cohort study. In addition to a standard cardiorespiratory monitoring, a simultaneous monitoring of abdominal and cerebral oxygenation was performed during the early neonatal period. The developed treatment guideline was based on a multisystem approach in diagnosis and optimization of the brain oxygen delivery and consumption. The proposed algorithm focused on maintaining an adequate perfusion pressure, cardiac output, oxygen, and carbon dioxide content in the arterial blood in newborns with respiratory disorders. Implementing the developed treatment guideline in the neonatal intensive care units should increase the treatment effectiveness in newborns with respiratory disorders. Implementation  developed  treatment  guideline  in  the  neonatal  intensive  care  units  should  increase  the  effectiveness of treatment in newborns with respiratory disorders.

Awareness of the importance of interpreting physiological data in context of each patient has led to a change the clinical paradigm from a "one size fits all" approach to a personalized one focused on achieving the optimal functional level of vital systems in individual infant. As a result, the problem facing many clinicians is the interpretation of this relatively new monitoring method. Both discrete and trend values are rich in information, however, in contrast to the well-known pulse oximetry, rSO 2 is determined by a wider range of physiological determinants [4]. In this context, correct interpretation of the NIRS-derived data, on the one hand will have fundamental importance for provide the appropriate interventions, and on the other, it will allow looking at the pathophysiology of disease from a different angle.
The aim of our study was to develop a guideline for dynamic monitoring of oxygen status based on rSO 2 data, as well as an effective therapeutic tool for newborns with respiratory distress.
Materials and research methods. From February 2013 to March 2016, we conducted this observational cohort study in our NICU in the National Scientific Practical Centre "Mother and Child". Seventy-eight newborns with respiratory disorders were enrolled (37 late preterm and 41 full-term). Recording the data began from the moment of birth (in the delivery room or operating room) and dynamically continued for the first 7 days of life.
Inclusion criteria: gestation age from 30 to 42 weeks; presence of respiratory distress requiring any respiratory support (oxygen therapy, non-invasive ventilation, conventional and high-frequency mechanical ventilation); arterial line for direct blood pressure and blood gases monitoring. Infants with congenital heart disease, multiple congenital malformations, history of sepsis, and surgical pathology were excluded.
In addition to standard cardiorespiratory monitoring, a simultaneous monitoring of abdominal and cerebral oxygenation were performed by the "INVOS 5100C" (Covidien, Medtronic, USA) with the Pediatric SomaSensor (SAFB-SM).
Unlike pulse oximetry (SpO 2 ), NIRS expresses full tissue hemoglobin oxygenation without subtraction of non-arterial data (venous and capillary). Thus, NIRS reflects the ratio of oxygenated to total hemoglobin for the mixed arterial, capillary, and venous vascular compartment underlying a given probe [1]. Regional oxygenation reflect the oxygen saturation in the venous (70 to 80 %), arterial (20 to 25 %) and capillary (5 %) [5]. Hence, as NIRS provides mainly information about hemoglobin oxygenation in post-capillary tissues, regional oxygenation can be considered as surrogate measures of the local oxygen utilization.
To assess cerebral oxygenation (crSO 2 ), the NIRS sensor was placed on the forehead, in the case of abdominal oxygenation (abdSO 2 ) -on the midline of the abdomen, above the pubic region (Fig. 1). All staff members attending were trained to use the device.
To better understand the balance between brain oxygen delivery and oxygen consumption, we used the cFTOE (cerebral fractional tissue oxygen extraction) then can be calculated from crSO 2 and arterial oxygen saturation: (SpO 2 -crSO 2 )/SpO 2 .
From clinical experience, cFTOE, calculated on the basis of NIRS and pulse oximetry data, showed a significant correlation with FTOE calculated during direct (invasive) measurement of oxygen content in venous and arterial blood in neonatal piglet model [6]. Thus, an increase in FTOE suggests an increase in the oxygen extraction by the tissues due to higher oxygen consumption in relation to oxygen delivery, while a decrease suggests less oxygen use compared to the supply. This also relatively compensates for low arterial oxygen content, as is often the case in newborns with lung disease or congenital heart defects [7].
In addition to the aforementioned NIRS data, we also analyzed the following: crSO 2 min is the daily minimal crSO 2 , crSO 2 max is the daily maximum crSO 2 , ∆crSO 2 is the daily amplitude of cerebral regional oxygen saturation (calculated as the difference between crSO 2 max and crSO 2 min), abdSO 2 min is daily minimal abdSO 2 , abdSO 2 max is the daily maximum abdSO 2 , ∆abdSO 2 is the daily amplitude of abdominal regional oxygen saturation (calculated as the difference between abdSO 2 max and abdSO 2 min).
To minimize the impact of third-party factors monitoring was carried out during calm patient state (without sedation), outside the phototherapy period and began at least one hour after enteral feeding [2,8].
Respiratory and hemodynamic support was carried out in accordance with the local NICU guidelines and the clinical protocols for intensive care in neonatology (order of the Ministry of Health of the Republic of Belarus dated January 28, 2011, No. 81).
Evaluation of blood gas parameters was performed with "АВL -835 Fleх" (Radiometer, Denmark). After collection arterial blood samples were immediately injected into a co-oximeter. We evaluated: PaO 2 -oxygen tension, mmHg, ctO 2 -the total content of oxygen, ml/dl, Lactate -lactate concentration, mmol/l, pH -the negative logarithm of the hydrogen ion activity, PaCO 2 -carbon dioxide tension, mmHg, p50 -oxygen tension at 50 % saturation of blood, mmHg. Paired assessments of blood gas parameters and regional oxygenation undertaken whenever an ABG was conducted as part of necessary patient care.
Echocardiography (ECHO) measurements were performed using commercially available Philips HD 11 XE (USA) in M-, B-, Doppler modes and in tissue Doppler imaging mode. For quantitative assessment of left ventricular function were used the common echocardiographic parameters: fraction shortening (FS), ejection fraction (EF). Pulmonary hypertension was defined based on the ratio of pulmonary artery pressure to systemic blood flow. Assessment of pulmonary artery pressure was carried out by tricuspid regurgitation peak velocity or by Kitabatake's method.
Data are presented as mean values ± SD for normally distributed continuous variables and median (interquartile range) when the distribution was skewed. Differences between the two groups were assessed using Mann-Whitney U-test for non-parametric data, Student's t-test for parametric data or χ 2 -test for categorical measurements. A correlation analysis was performed to investigate if there was an association between between regional oxygenation and red blood cell parameters, as well as arterial blood gas data. Correlations were performed using Spearman's rank correlation coefficient (ρ) or Pearson's correlation where appropriate. The pre-and post-transfusion values of all measurements were compared using Wilcoxon matched pairs test. To identify physiological determinants of cerebral oxegenation and cFTOE simultaneous multiple linear regression was performed. Checking for multicollinearity before building a multiple regression model was carried out based on the analysis of the Variance Inflation Factor and the correlation matrix. To calculate the probability that a stabilization of the cardiorespiratory status (weaning from inotropes/ vasopressors and mechanical ventilation) will occur at certain points in time, the Kaplan-Meier method was used. To compare groups a log-rank test was performed. p < 0.05 was considered statistically significant. The statistical analyses were performed using MedCalc 19.1 (MedCalc Software Ltd., Ostend, Belgium).
Results and its discussion. In this study 78 newborn infants with respiratory disorder requiring respiratory support (non-invasive, conventional or high-frequency oscillatory lung ventilation) were included. According to the tasks of the study, clinical and statistical hypotheses were made. The flow chart of the study was developed (Fig. 2).
Condition of all included newborns considered as severe because of profound pulmonary insufficiency with hemodynamic instability. All patients were provided with central venous catheter for inotropic and vasopressor support. Clinical characteristics of patients are outlined in the Table. First we checked the assumption that regional oxygenation's data could be used as a tool for estimation the effectiveness of respiratory and hemodynamic support. We revealed that ∆crSO 2 has statistically significant differences depending of initial clinical status at the beginning of intensive care and in the period of further stabilization. Patients with more severe clinical status had more statistically significant daily shifts of crSO 2 (p < 0.0001). Patients who was stabilized (weaning from inotropes/vasopressors and mechanical Fig. 2. Study flow chart ventilation) to 96 hours of life had ∆crSO 2 lower than those, who was stabilized to 7 days of life as well as those who did not reach hemodynamic stability in early neonatal period 8 (5; 15) % vs 11 (8; 14) and 13 (9; 22) % respectively. Moreover, we found that premature infants with ∆crSO 2 ≤ 20 % in the first day of life had probability of reaching cardio-pulmonary stability in early neonatal period 3.3 times higher (95 % CI (1.5-6.9), p log-rank = 0.0023) compared with patients with ∆crSO 2 more than 20 % during the 1st day of life.
Both very low and very high values of daily crSO 2 could contribute to the increase of ∆crSO 2 . As target we considered crSO 2 and сFTOE values 78 ± 7 % and 0.22 ± 0.07 which are stated to be "physiologic" according the literature and confirmed by us during pilot study [9,10].
Сerebral oxygenation levels laying out of physiologic values give us signs of imbalance between oxygen delivery and consumption so multisystem assessment of patient is needed (Fig. 3).
Estimation of hemodynamic status. Cerebral autoregulation is known to be a mechanism that allows to mitigate effects of systemic pressure fluctuations on brain blood flow. That's why crSO 2 levels below physiologic range in patient with systemic hypotension are most likely to indicate loss of auto-regulation and require immediate intervention to increase cardiac output and tissue perfusion. As we established, the negative relationship between the blood lactate level and the value of the crSO 2 , as well as the positive one with cFTOE indirectly confirm this fact [11]. Increase of lactate (traditional marker of hypoperfusion and tissue hypoxia) is accompanied, under equal conditions, by a reduction in the supply of oxygen to tissues and an increase in O 2 extraction. As a results, we will see certain crSO 2 changes reflecting impaired oxygen status of the brain. As the first markers of systemic hypoperfusion we recommend to use indicators widely available in everyday clinical practice: prolonged capillary refill time, oliguria and arterial lactate.
According to our data, the SCOR coefficient is a useful clinical tool to evaluate end-organ hypoperfusion. As we found SCOR below 0.80 indicated hypoperfusion with a sensitivity of 53.4 % and a specificity of 85.9 % [12].
Bedside echocardiography is an integral part of the modern approach in assessing hemodynamic status in newborns [13]. But ECHO data also add useful information about the effectiveness of oxygen delivery to the brain. In particular, our study found that the systolic function of left ventricle was positively  It has been proved that the use of both dopamine and epinephrine for early systemic hypotension in newborn improves brain perfusion and oxygenation [14]. Noteworthy that vasopressors use (epinephrine and norepinephrine) for hemodynamic instability in newborns increase preload because of α-adrenoreceptor mediated peripheral vasoconstriction. As a result, heart contractility (and eventually brain perfusion) could be compromised so echocardiographic assessment of such patients is required. As our study shows, newborns received combination: dopamine ≥10 mcg/kg·min and epinephrine (norepinephrine) ≥0.05 mcg/kg·min, had highest level mean arterial pressure and at the same time lowest level of crSO 2 and closed to minimal indicators of left ventricle's function (EF and FS) [15]. In addition, the use of epinephrine and norepinephrine in newborn infants, even in case of reduced peripheral vascular resistance (dopaminerefractory distributive shock), may involve adverse changes in the oxygen status of the brain [16]. Thus, in our view, diminishing the dose of vasopressor (if already used) in case of low cardiac output and decreased cerebral oxygenation seems a reasonable tactic.

Main clinical characteristics of studied patients
Evaluation of blood oxygen transport. We found statistically significant correlations between crSO 2 and cFTOE on the one hand, and erythrocyte count, hemoglobin, and hematocrit on the other, confirming the statement that crSO 2 depends on the appropriate delivery of oxygen to the brain [11]. We have shown that tissue oxygenation impairment in newborns with respiratory distress and anemia manifests not only with such invasive criteria like decrease of systolic blood pressure or arterial lactate elevation but also non-invasive indicators like falling of crSO 2 and increase of cFTOE.
It is known that NIRS data tended to return to normal after the correction of the anemia [17]. It has also been found that a newborn with a cFTOE > 0.22 is 3.5 times more likely to have anemia requiring blood transfusion [17]. Additionally, as we have shown earlier, correction of hemoglobin by blood transfusion in full-term newborns with hemolytic anemia also was accompanied by increase of crSO 2 [18]. So it is clear that NIRS can be used in NICU not only for early diagnosis of tissue oxygen imbalance because of anemia, but also for monitoring of the effectiveness of blood transfusion.
Evaluation of respiratory status. It is widely known now that PaCO 2 is a major determinant of cerebral perfusion and consequently oxygen status of the brain [19]. Significant positive correlation between PaCO 2 and crSO 2 also confirms this statement [11,15]. It is extremely important in case of mechanical ventilation in newborns because it could be accompanied with inadvertent hyperventilation and hyperoxia [15]. For example, we established by ROC-analysis that crSO 2 ≤ 66 % in infant on ventilation and combined inotropic support may indicate the presence of hypocapnia with sensitivity of 40.0 % and specificity of 100 % (AUC = 0.670, p = 0.0343) [15]. Thus, in our opinion, the reduction of the minute volume of ventilation is justified not only in the case of hypocapnia, but also in clinical situation when PaCO 2 approaches the lower limits and simultaneously the crSO 2 is below physiologic range.
Lung ventilation has negative influence on systemic hemodynamics and brain blood flow because of increase of intrathoracic pressure reducing preload (and eventually cardiac output) and hampering venous return from the brain [20]. Our previous study about predictors of crSO 2 in newborns on respiratory support revealed that crSO 2 was inversely correlated with mean airway pressure [15]. This means that when patient on ventilation have a downward crSO 2 trend (especially when left heart contractility decreased) lung emphysema should be excluded and MAP should be diminished.
Until recently, pulse oximetry was the only effective method of controlling oxygen supply to meet metabolic needs. Despite that, monitoring SpO 2 solely do not provide the information about adequacy of oxygen supply of the brain. As widely known, the most typical response on the desaturation in newborn patients is giving an additional oxygen. Almost every time when stable desaturation occurs the fraction of oxygen received by patient is increased by 10-20 % or more until SpO 2 is getting to be in physiological range.
Studies of desaturation episodes mainly focus on the harmful effects of hypoxia and hypoxiaischemia [21]. However, the oxygen supply to the central nervous system can reach potentially dangerous levels in the post-hypoxic reperfusion, especially in patients with hyperoxia, which, according to our data, is typical for patients with the most compromised cardio-therapy status [15].
However, according to the results of our research, there were no statistically significant relationship between the PaO 2 and saturation of the arterial blood and crSO 2 , which is consistent with the observations of C. Hunter [11,22]. But it is necessary to note that if SaO 2 >90 %, and PaO 2 exceeds 80 mmHg oxyhemoglobin dissociation curve becomes relatively flat and a further increase of PaO 2 has a relatively minor effect on oxygen saturation of the blood. Since study participants had median values of SaO 2 and PaO 2 significantly above the physiologic range, this may partly explain the lack of a statistically significant relationship.
Another one reason for above physiological crSO 2 is a low oxygen extraction in the brain, in front of its adequate delivery (severe hypoxic-ischemic encephalopathy, profound anesthesia) [2,23].
Considering all of the above, if patient on respiratory support has stable cerebral hyperoxia the first step to consider is the lowering oxygen fraction not only in cases when SpO 2 is above physiologic range but when it closer to them. The vasodilation effect of hypercarbia should also be taken into account, so in the case of CO 2 biuld-up with an accompanying cerebral hyperoxia, an increase of minute ventilation is justified as an appropriate step.
Our novel diagnostic and treatment guideline is based on integrated estimation and optimization of brain's oxygen supply and consumption via maintaining proper perfusion pressure, cardiac output, O 2 and CO 2 arterial content in newborns with respiratory distress.
In our research total oxygen consumption of the newborn's brain was not measured directly. Thus, all conclusions about oxygen extraction are based on a reasonable assumption, about a fairly stable O 2 metabolism in the brain.
We determined minor differences of crSO 2 and cFTOE in term and late pre-term newborns but physiological determinants of brain's oxygen status were independent of the gestational age [15,27]. Thus, our approach is applicable to both term and late-term newborns with respiratory disorders in early neonatal period.
It is also important to note that routine manipulations performed in the NICU, such as changes in body position, repositioning or aspiration from the endotracheal tube, withdrawal of blood from umbilical catheters, may adversely affect cerebral hemodynamics and oxygenation [14,28,29]. As these changes are usually temporary, they have not been taken into account.
As oxygen status of the brain of a newborn child with respiratory disorders is a continuum and almost all patients have a combination of several reasons for imbalance between oxygen supply and consumption of oxygen by the brain, selection of separate components in proposed program of stabilization of cerebral oxygen status is highly conditional.
If crSO 2 lays out the specified physiological range and intervention is needed we recommend stepby-step approach: only one action at time, then re-evaluation in 30-60 min and another action.
It is very important to have an ECHO available to assess patient's hemodynamics.
Our integral approach to monitoring of oxygen status with dynamic control of regional oxygenation as well as complex of effective interventions allow clinicians to be more effective in management of newborn patients with respiratory disorders.
Conflict of interests. The authors declare no conflict of interests.