Predischarge cerebral oxygenation and psychomotor outcome in very preterm infants: is there an association?

This observational study aimed to investigate whether predischarge cerebral oxygenation (CrSO2), monitored by near-infrared spectroscopy, correlates with later psychomotor outcome in very preterm infants. Infants <32 weeks’ gestation or <1500 g without evidence of major brain lesions underwent a 3-h continuous CrSO2 monitoring before hospital discharge. Psychomotor development was assessed at 6, 12, 18, and 24 months using the Griffiths Mental Developmental Scales. The developmental quotients (DQ) at each follow-up appointment were correlated with predischarge CrSO2. Significant correlations were adjusted for possible confounders. Sixty-three infants were enrolled. A significant correlation between CrSO2 and DQ was observed at 6 months ca (p=0.010), but not at later psychomotor assessments. This correlation was confirmed significant (b=0.274, p=0.038) even after the adjustment for relevant covariates. Conclusion: According to these preliminary findings, the association between predischarge CrSO2 and psychomotor development over the first 24 months in preterm infants without major brain lesions is time-limited. Hence, this parameter may not represent an effective predictor for medium-term neurodevelopment. What is Known: • Prematurity is a major risk factor for adverse neurodevelopment. • The validation of clinical tools for psychomotor outcome prediction may aid to identify high-risk preterm infants who might benefit from early interventions. What is New: • In infants without major brain lesions, predischarge CrSO2 correlates with psychomotor outcome at 6 months ca but not later, indicating a short time predictability.


Introduction
Despite recent improvements in neonatal care, preterm birth is still burdened by a significant risk of neurocognitive disabilities [1]. Due to the intrinsic neuroplasticity that characterizes early brain development, a timely establishment of supportive interventions may counteract the functional constraints related to prematurity [2]. The validation of predicting tools for neurodevelopmental outcomes might help to identify preterm neonates who could benefit from these habilitative interventions. Near-infrared spectroscopy (NIRS) provides a non-invasive continuous monitoring of cerebral oxygenation (CrSO 2 ). Evidence on the possible association between CrSO 2 and neurodevelopment in preterm infants is limited to early postnatal phases [3][4][5], when transitional hemodynamic instability may significantly affect CrSO 2 .
We aimed to explore whether predischarge CrSO 2 , monitored in clinically stable preterm infants without evidence of major brain lesions, correlates with psychomotor outcome over the first 24 months.

Materials and methods
Preterm infants with a gestational age (GA) <32 weeks or birth weight (BW) <1500 g admitted to the Neonatal Intensive Care Unit of S. Orsola-Malpighi University Hospital, Bologna, Italy, between February 2013 and June 2014 were consecutively enrolled in this observational study. Infants with major congenital abnormalities, hypoxic-ischemic encephalopathy, or neuroradiological evidence of major brain lesions (i.e., periventricular leukomalacia, intraventricular hemorrhage grade >II, or post-hemorrhagic ventricular dilatation) were excluded. Due to its influence on CrSO 2 , a hematocrit <30% at the time of NIRS monitoring was also considered an exclusion criterion. The study was conducted in conformity with the principles and regulations of the Helsinki Declaration. Approval was granted by the Ethics Committee of S. Orsola-Malpighi Hospital, Bologna, Italy (ref. EM 236/2016/U). Written, informed consent to participate in the study was obtained from the infants' parents. CrSO 2 was monitored using an INVOS 5100 oximeter (Medtronic, Boulder, CO, USA) when the infants met the following criteria for hospital discharge [6]: physiologic stability, self-ventilating in air, stable cardiorespiratory function (i.e., no need for cardioactive drugs), full enteral feeding (160 ml/kg/day), sustained weight gain. The NIRS sensor was placed on the central forehead and maintained in the same position for the whole monitoring period.
NIRS recording was started after at least 30 min of stable signal and continued for 3 h. During this period, handling procedures were minimized and feeds were administered prior to or after the monitoring. After the recording, NIRS traces were visually inspected and artifacts were removed. CrSO 2 values were averaged over 3 h and used for statistical analysis. The coefficient of CrSO 2 variability, which is defined as the ratio of the standard deviation (SD) to the mean of a given set of measurements and provides an estimation of the signal stability [7], was also calculated.
The following clinical data were retrieved: intrauterine growth restriction (IUGR, defined as a fetal growth reduction resulting in BW<10 th percentile for GA), sepsis development (defined as positive blood culture in the presence of clinical signs of infection), bronchopulmonary dysplasia (defined by the need for supplemental oxygen at 36 weeks' postmenstrual age), and maternal education level (basic education, high school degree, university/master's degree).
At 6-, 12-, 18-, and 24-month corrected age (CA), the enrolled infants were included in the local neurodevelopmental follow-up program performed by means of the Griffiths Mental Development Scales 0-2 years [8]. These scales investigate five developmental areas (locomotor, personal and social skills, hearing and language, eye and hand coordination, performance) and, based on the quotients for each of the aforementioned sub-scales, provide a general developmental quotient (DQ) of the infant's abilities. The scales were administered by a trained neuropsychologist, blind to the NIRS data. The quotients were calculated using the tables of standardized scores, corrected for GA. Based on the standard deviations (SD) of DQ, psychomotor impairment was defined as mild (−1 to −2 SD), moderate (−2 to −3 SD), and severe (<−3 SD) [8].

Statistical analysis
A sample size of 62 subjects was calculated for a twotailed alpha of .05 and power of 80% to detect a correlation coefficient ≥0.35. Data distribution was evaluated by the Shapiro-Wilk test. Numerical variables were summarized as mean±SD or median (interquartile range) as appropriate; categorical variables were summarized as frequencies and percentages. Since both mean CrSO 2 values and DQ scores followed a normal distribution, the correlation between CrSO 2 and DQ at each follow-up appointment was analyzed with Pearson's test. A hierarchical multiple regression model was built to adjust the observed results for covariates known to influence preterm infants' neurodevelopment. Statistical analysis was performed with IBM SPSS Statistics version 25.0. The significance level was set at p<0.05.

Results
Sixty-three infants were included in this study. The clinical characteristics of the study population are summarized in Table 1. The mean postmenstrual age at NIRS monitoring was 34.8±2.3 weeks. Mean predischarge CrSO 2 was 71.1±7.2%, with a mean coefficient of variation of 0.04±0.01. Only 5 (8%) infants had mean CrSO 2 values <55%.

Discussion
According to the present results, predischarge CrSO 2 in preterm infants with no evidence of overt brain lesions significantly correlates with DQ at 6 months CA, but not later.
Prematurity is associated with typical changes in brain structures and connectivity, with long-term effects on cognitive, social, and language skills [9][10][11]; hence, predicting neurodevelopmental outcomes in the preterm population would aid to reduce the ensuing burden of disability.
To date, the role of early cerebral NIRS monitoring in predicting neurodevelopment has been investigated only in a few studies, which reported a significant association between low CrSO 2 and adverse outcomes at 15 and 24 months CA [3][4][5], while in the present study, this association was limited to earlier assessments. However, several methodological differences with these studies need to be acknowledged. First, NIRS monitoring was performed in different time periods: Verhagen et al. [4] assessed CrSO 2 on days 1 to 5, 8, and 15, while Cerbo et al. [3] and Alderliesten et al. [5] carried out a continuous CrSO 2 monitoring over the first 48 and 72 h, respectively. During early postnatal phases, multiple conditions (e.g., patent ductus arteriosus, hypotension, mechanical ventilation) can alter preterm infants' hemodynamics, resulting in a reduced CrSO 2 and increasing the risk of brain injury. However, early CrSO 2 is blind to the effect of several prematurity-related complications that can later occur during hospital stay (e.g., bronchopulmonary dysplasia, sepsis, necrotizing enterocolitis) and are known to affect preterm infants' neurodevelopment. By monitoring CrSO 2 before discharge in the most stable clinical conditions, we aimed to obtain a snapshot of the status of brain parenchyma net of the occurrence of previous complications. Moreover, the development of a predictive tool for psychomotor outcome in the predischarge phase may have potentially useful implications to plan taking-into-care and supportive interventions by the community services for at-risk infants.
Differently from the previous studies, we chose to exclude infants with evidence of brain lesions, which can affect CrSO 2 measurements and are associated with adverse neurodevelopment [12]. Depending on local resource allocation policies, preterm infants without such major risk factors as overt brain damage may not be included in specific followup programs, with the risk of missing possible impairments; hence, a low-cost and non-invasive tool for neurodevelopment prediction in this specific population may have useful clinical implications. However, the exclusion of preterm neonates with major brain injury from the present study may have contributed to the lack of medium-and long-term predictivity of CrSO 2 on neurodevelopment.
The following study limitations need to be acknowledged. The small number of infants with abnormal DQ did not allow to split the study population according to the degree of psychomotor impairment. Similarly, the low percentage of infants with CrSO 2 values <55% precluded to stratify NIRS data according to this threshold adopted in early postnatal studies [4,5]. The 3-h timescale adopted for NIRS measurements may be relatively short to provide a consistent CrSO 2 assessment; however, the stable clinical and hemodynamic conditions in the predischarge phase and the minimal CrSO 2 variability support the reliability of this monitoring window. Although predischarge CrSO 2 correlates with psychomotor development at 6-month CA, we failed to demonstrate a medium-term association, consistently with the several factors that may contribute to influence neurodevelopment at increasing ages. Given the short-term correlation observed in the present study, whether CrSO 2 monitoring at each follow-up evaluation may be useful to predict preterm infants' neurodevelopment from time to time, may warrant being explored.