Early post-operative neurodevelopment and visual assessment in neonates with congenital heart disease undergoing cardiac surgery

Assessment of neurobehavior and visual function of newborns with congenital heart disease during the post-operative period to identify infants at risk of neurodevelopmental and visual impairment. Prospective study that included 45 newborns who underwent cardiac surgery. Newborn Behavioral Observations test (NBO) and “ML Battery of Optotypes” were used for assessment. The median NBO global score was 2.4 [2.1–2.6]. Total days of oral morphine [p = 0.005] and total days of sedation [p = 0.009] were strongly related to abnormal evaluations. Time of cerebral regional oxygen saturation (CrSO2) under 40% during surgery and increased lactate were related to abnormal motor evaluation. Only 14.5% of patients presented pathological results in visual evaluation. We have demonstrated alterations in attention, autonomic, motor, and oral motor function. Duration of sedative medication, time of CrSO2 under 40% during surgery, and increased lactate are the most important risk factors. No significant visual impairment was detected.


INTRODUCTION
Newborns with congenital heart disease (CHD) undergoing early cardiac surgery (CS) are at increased risk for brain injury and subsequent neurodevelopment deficits. Cognitive and motor impairment as well as behavioral and learning disabilities are some of the main issues reported in this population [1][2][3]. Furthermore, there is also evidence of visual impairment. Ocular abnormalities have been reported in up to 32.5% of neonates with cyanotic heart lesions or coarctation of the aorta, particularly in the form of retinal vasculature changes [4,5].
There is an increasing evidence that fetuses with CHD present with abnormal brain maturation [6]. Their brain development appears to be delayed when compared to healthy newborns. Interestingly, neurological alterations reported in this population are similar to those detected in preterm babies, with a high risk for motor and cognitive delay [7]. Moreover, cardiac surgery has been suggested to have a negative impact on brain maturation, especially in the youngest patients [8]. Inadequate tissue perfusion and oxygenation during this period has been associated with poor prognosis [9].
Based on these factors, early detection and early intervention will be essential to minimize suboptimal neurodevelopment and improve prognosis.
Many methods for early evaluation have been validated in order to improve prognostication of neurodevelopmental problems during the neonatal period [10][11][12]. One of the frequently used tools is The Neonatal Intensive Care Unit Network Neurobehavioral Scale (NNNS), a standardized assessment developed to identify and quantify neurobehavioral abnormalities in high-risk newborns, including newborns with CHD [13]. Lower NNNS scores have been reported in this population compared to healthy infants, reflecting suboptimal performance in neurobehavioral domains [10]. The Newborn Behavioral Observations system, a family-centered, relationship-based tool, was designed to sensitize parents to their baby's competencies and individuality, and has been used to foster positive parent-infant interactions from the very beginning [14]. This test consists of 18 neurobehavioral observations of the infant when sleeping, awake or crying. The domains (or subscales) are: autonomic (respiration, digestion, skin color), motor regulation (tone, movement, postures), organization of state (range, robustness, transition patterns), and responsiveness (robustness, transitions).
In addition, early detection of visual disorders, closely related to suboptimal neurodevelopment, is of paramount importance in order to minimize adverse visual and long-term sequelae. While visual evaluation is often a component of neurobehavioral tests, specific examinations should be separately conducted to detect more subtle problems. Leonhardt et al. reported the use of the "ML Battery of Optotypes" to assess visual function, especially in low-risk preterm newborns [15][16][17]. Its use in the CHD population has not been previously described.
The aim of this study is to evaluate neurobehavioral patterns and visual function of newborns with CHD during the immediate post-operative period to identify those infants at increased risk for future neurodevelopmental and visual impairment.

MATERIALS/SUBJECTS AND METHODS
This was a prospective, observational study including neonates less than 30 days old undergoing heart surgery from December 2017 to November 2020 in our center. Exclusion criteria were a known or clinically suspected genetic syndrome, history of birth asphyxia, or preexisting brain damage. The study was reviewed and approved by the institution's Ethics Committee (PIC 120- 17) and was performed in accordance with the Declaration of Helsinki. Written informed consent was obtained from the parents or legal guardians of all participants.
Demographic and clinical data Demographic and clinical data including gestational age, birth weight, head circumference, gender, prenatal diagnosis, acidosis, length of stay (LOS), age at surgery, use of cardiopulmonary bypass (CPB) and/or deep hypothermic circulatory arrest (DHCA), minimum central temperature during surgery, postoperative peak serum lactate, total days of use of sedative agents, and age at evaluation were collected.
Each patient was assigned to one of the following four diagnostic cardiac groups [18,19]: Group I: two-ventricle CHD without aortic obstruction; Group II: two-ventricle CHD with aortic obstruction; Groups III: single-ventricle CHD without aortic obstruction; Groups IV: single-ventricle CHD with aortic obstruction.

Operative management
Anesthetic management followed institutional cardiac anesthesia protocols that included high-dose fentanyl, inhaled isoflurane and muscle relaxants. Neither benzodiazepines nor barbiturates were administered during surgery. Target temperature during CPB was 22-34°C. Deep hypothermia (14.1-20°C) was used in cases requiring circulatory arrest [20]. Selective anterograde cerebral perfusion was not performed.
Cerebral regional oxygen saturation CrSO 2 was measured with near-infrared spectroscopy (INVOS 5100C Cerebral/Somatic Oximeter, Medtronic, Minneapolis, MN). Two appropriately sized sensors were placed on the patient's forehead during surgery and remained in place up to 72 h after surgery. CrSO 2 levels were continuously recorded. Data were exported in 30-second interval averages. Time in minutes spent under 40% CrSO 2 , under 50% CrSO 2 , outside of the range 50-70% CrSO 2 , and over 85% CrSO 2 was also analyzed.

Amplitude-Integrated Electroencephalogram (aEEG)
Continuous aEEG (Nicolet One TM; Natus, Middleton, WI and Olympic Brainz Monitor; Natus, Seattle, Washington) was used to monitor brain electrical activity during surgery and up to 72 h after surgery. This device recorded a 2-channel EEG using signal acquired from two central and two parietal hydrogel electrodes (C3, C4, P3, and P4). Impedance below 20Ω was assured. Two neonatologists trained in aEEG interpretation and blind to patient information reviewed tracings, analyzing background patterns in accordance to the previously described Hellström-Westas classifications [21]. Presence of seizures as well as the total duration of seizures in minutes were recorded.

Post-operative management
All patients received milrinone combined with dopamine and epinephrine as needed. Postoperative analgesia and sedation were achieved with continuous infusions of either morphine or fentanyl and dexmedetomidine. Morphine sulfate and clonidine were administered to patients who later developed symptoms of withdrawal.

Newborn behavioral observations and visual function
Patients were evaluated at least five days after surgery, when they were either off or receiving low-dose sedative drugs, and always before being 90 days old. Both tests were administered at the same time when possible.
The NBO was performed in all study participants by a certified neuropsychologist. Data were analyzed globally and by subdomains: Autonomic, Motor regulation, Organization of State, and Responsiveness. Scoring consisted of 1 for poor outcomes, 2 for intermediate outcomes, and 3 for normal outcomes.
Visual behavioral responses of alertness, fixation, attention and tracking were analyzed using the "ML Battery of Optotypes". This test involves eight high-contrast stimuli that are shown consecutively, one optotype at a time, at a distance of 15-20 cm in front of the newborn's eyes. Optical interactions, optical function, optical perception and ocular fields were described and quantified [15,17].
The results of the assessment were grouped in different functional areas that included: (i) visual function (measurement of reflex response, optotype follow-up, eyes movement, pupillary reaction); (ii) optical function (pupillary reaction, targeting, fixation, visual axis); (iii) optical perceptual function (reflex response, alert, attention); (iv) ocular motility and visual fields (eyes movement, optotype follow-up) and (v) optical perceptual interactive function (human face response, optotypes response, color response) (supplementary material).

Statistical analysis
Study data were stored using the institution's Research Electronic Data Capture (REDCap) [22]. Continuous variables were expressed as median and interquartile range for non-normally distributed values. The U-Mann Whitney test was used to compare independent measures in non-parametric continuous variables. Chi square was used to compare dichotomous variables. Main variables were classified as positive or negative results. Univariable logistic regression was performed to evaluate the effect of each variable individually. Multivariable analysis with adjustment for age at time of evaluation was performed to identify risk factors for pathological results. Multivariable regression was also performed to account for possible confounders. Results were considered significant if p < 0.05. Analysis was conducted using statistical package STATA.13 (College Station, TX).

Population characteristics
Forty-five infants who underwent CS during their first month of life between December 2017 and November 2020 were included. Demographic characteristics and clinical data of all patients are described in Table 1. Main results of peri-operative CrSO2 are included in Table 2. Type of surgery is described in Fig. 1.

Neurodevelopmental assessment
Neurodevelopmental assessment was conducted on median day of life 31.2   Fig. 2A).
Global NBO results were analyzed according to gender (p = 0.8092), prematurity (p = 0.3349), DHCA duration (p = 0.8731), need for CPB surgery (p = 0.1467) and presence of seizures during surgery (p = 0.1180), and none of these variables seem to be related to nor have an influence on the results.
The main results of each domain and subdomain score are included in Table 3. This table includes clinical presentation, score description, percentage of each score (included as pathological score/suboptimal score /normal score), and percentage of pathological results for each domain of evaluation. Results were expressed as percentages.
All clinical variables were analyzed in relation to neurodevelopmental outcomes. No risk factor was related to global NBO score. Each risk factor was further evaluated for influence on individual domains.

State organization
In the sub-domain of state organization, 87% of patients had normal visual habituation, but only 57% had normal auditory habituation. Days of sedation was the most impactful clinical factor, as total days of oral morphine (OR:1.1; [1.01-1.3]) and total days of sedation (OR:1.13; [1-1.3]) were most strongly related to abnormal evaluations.

Motor organization
Ninety-five percent of infants had appropriate tone and strength in upper and lower extremities, but low shoulder and neck tone was seen in 58% of the cohort. Regarding motor skills related to feeding, 18 Fig. 2D]. No relationship was identified between motor organization and days of sedation.

Attention organization
All infants displayed an appropriate reaction to faces and voice and a correct orientation to voice. Specifically, infants attended more readily to face and voice than to a toy that made noise. In the subdomain of visual following, however, 12.2% of patients had an abnormal response. Patients with poorer outcomes in this domain spent more time with CrSO 2 under 40% during the immediate postoperative period (Fig. 2E), however, the difference did not persist when correcting for age at time of assessment (p = 0.125).

Visual Assessment
A global visual evaluation was also performed using ML optotypes. Each function was separately evaluated. Table 4 outlines clinical presentation, score description, percentage of each score, and percentage of pathological results of visual assessments. While no major disabilities in visual evaluation were detected, visual function was the sub-domain with the poorest results. A summary of the main results from the ML optotypes assessment is presented in Fig. 3. Main risk factors previously related to ocular abnormalities in CHD patients were analyzed, including cyanosis (p = 0.469), CPB surgery (p = 0.3142) and prematurity (p = 0.3659), and none of them were related to optical alterations in our population.

DISCUSSION
In our cohort of newborns with CHD undergoing cardiac surgery, prolonged use of opioids, increased lactate levels after surgery, and time of decreased CrSO 2 during surgery were the main risk factors for early suboptimal performance across neurobehavioral domains. Interestingly, though visual impairment has been reported in newborns with CHD, no major disabilities in this area were detected in our cohort.
Long-term neurocognitive impairment is one of the most prevalent issues affecting neonates with CHD. In these patients, central nervous system maturation at birth has been reported to be delayed approximately one month compared to healthy newborns, suggesting a prenatal origin of cerebral abnormalities [3,23]. Moreover, some of the events that happen during the perioperative period likely also have a negative impact on brain development [24]. This evidence supports the idea that brain injury in infants with CHD is the result of a complex interaction of patient-specific factors and environmental influences, including the effects of CS and perioperative care [25]. Accordingly, the etiology of neurodevelopmental problems has been considered multifactorial and has not been clearly elucidated [26], but some factors such as inadequate tissue perfusion and oxygenation have been associated with poor prognosis [9].
There is a lot of evidence about long-term neurodevelopment alterations in this population [27][28][29], but few studies have been able to report early neurological dysfunction prior to hospital discharge [10,12]. The adapted version of the NBO assessment used in this study has been previously used to assess neurodevelopment in the neonate undergoing CS [23]. Complex cardiac anatomy, surgical complexity, younger gestational age, extracardiac congenital anomalies, and the need for pre-operative cardiac catheterization have been associated with poor outcomes [10]. To our knowledge, our study is the first to analyze whether the use of neuromonitoring, serum biomarkers, or other perioperative variables can be used to predict abnormal neurobehavioral results using this tool. In our population, most patients displayed normal visual habituation, but nearly half did not achieve normal auditory habituation. Prolonged use of opioids was the main risk factor related to abnormal responses in this domain. Abnormalities in state organization have been previously described [10,30], but no risk factors have been associated with them. Postoperative pain management is especially difficult in newborn patients [31]. Most drugs given for pain control have a sedative effect that may influence the neurological assessment, and some autonomic functions could be impacted in those patients who require more days of sedative drugs. In this line, suboptimal neurobehavioral scores during the neonatal period and worse neurocognitive outcomes at 2 years of age were also reported in preterm infants who require sedation or in drug-exposed infants [32,33], supporting our results.
Limperopoulos and colleagues reported in a study with children undergoing CHD surgeries that motor impairments, as well as behavioral difficulties, were frequently observed at 20 months of age, and 23% of their cohort showed impairments across developmental domains [34].
Increased or decreased motor tone after surgery was also described in early evaluations [23]. More than half of our patients had abnormal axial tone. Increased time with low cerebral regional oxygen saturation (time of CrSO 2 under 40%) during surgery and high levels of lactate after surgery seem to be the main risks factors for early motor alterations. These findings are in line with those of Aly and colleagues, who reported correlations between decreased cerebral oxygenation (time of CrSO 2 under 40%) and increased lactate levels with poor neurological outcomes at one year [1].
Despite the fact that low attention and less interaction with the environment have been described in CHD patients [4], in our population, the attention scores were optimal, and most of the patients had a very good response to voice and faces. Suboptimal inanimate visual following, only detected in few patients, was also related to decreased cerebral oxygenation (time of CrSO 2 under 40%) during the postoperative period.
Massaro and colleges, using the Neonatal Intensive Care Unit Network Neurobehavioral Scale, another standardized test developed to identify and quantify neurobehavioral abnormalities in high-risk newborns, reported vulnerabilities in regulation and stress domain scores [10]. In their study, CHD patients were compared to healthy newborns and were evaluated before and after surgery. In our population of post-surgical patients, only 5% presented signs of stress and diminished capacity to respond to stimulation, and less than 10% were inconsolable while crying. No risk factors were associated with these domains in our study.
Prolonged use of opioids, decreased cerebral oxygenation (time of CrSO 2 under 40%), and increased lactate levels after surgery were the main risk factors for early neurological dysfunction in our patients. In the literature, male sex, prematurity, and more than 40 minutes of DHCA have been associated with worse neurological outcomes. We did not find this relationship in our population, however, even after a regression analysis correction, perhaps due to sample size [10]. Use of CPB, type of CS, or presence of seizures during surgery were other risk factors analyzed, but none of them were related to poor outcomes in early neurological evaluation.
The assessment of visual function has become part of any neurological examination. The ML Optotype Battery analyzes the visual behavioral responses of alertness, fixation, attention, and tracking, as well as global visual ability. This protocol has been validated in healthy infants and in preterm patients [17]. One-third of patients with CHDs not linked to genetic syndromes have been reported in the literature to have ocular abnormalities. These findings were more prevalent in patients with cyanotic CHD [5].  Despite this existing association, to our knowledge, this is the first study to include a specific evaluation of visual function in newborns with CHD after surgery. Most of our patients had a normal visual assessment, however, visual function was the subdomain with the worst results. These patients had a low or very low measurement of reflex response, optotype follow-up, eye movement, and pupillary reaction. Seven percent of patients had a pathological optical perceptual function, with low or very low reflex response, alertness and attention.
As we have previously mentioned, ocular alterations have been related to cyanotic CHD [5], but it has also been reported that these abnormalities may regress after surgery. In the same line, there are some studies that suggest the presence of retinal alterations in patients with aortic coarctation [4], though we did not find a correlation between abnormal visual assessment and cyanosis or aortic coarctation.
Our study has some limitations. First of all, we had a relatively small sample size which can decrease the strength of associations. Moreover, we did not have a control group of healthy patients with whom to compare findings. Furthermore, our evaluations only occurred in the post-operative period. While this timing can be helpful for early prognostication, it does not provide insight into persistence of abnormal results. In order to assess how early findings relate to longer-term outcomes, these children have been included in a follow-up program with systematic evaluations at 12 and 24 months.
In conclusion, we have demonstrated alterations in the interrelated areas of autonomic function, motor function, oral motor function, and attention in infants recovering from cardiac surgery. Duration of sedative medication, time of CrSO 2 under 40% during surgery and increased lactate are the most important risk factors related to poor outcomes in our population. It is also important to note that no significant visual impairment was detected in our population.
Neurodevelopmental assessment of the neonate using the NBO in the post-operative period may provide useful information about early neurological status of CHD patients. Early detection of neurologic or visual impairments is of paramount importance to prioritize early intervention, in order to prevent the long-term sequelae of suboptimal neurodevelopment. Correlation of these early findings with later follow-up is essential to determine its clinical significance.

DATA AVAILABILITY
All relevant data are included in the manuscript. Data are available on reasonable request.