Preterm birth is a major paediatric public health problem as prematurity is associated with a considerable risk to develop cognitive, behavioral, neurosensory, and motor disabilities. The rates of preterm birth have risen up in many European countries,1 and a concomitant successful improvement in survival has been reported through the years.2 However, a missing reduction in the prevalence of severe disability still characterizes the extremely prematurely-born population,2 being the gestational age the major discriminant for a good-quality survival of these extremely preterm babies:2-4 the lower the gestational age, the higher the risk of death and neurodevelopmental impairment.
Brain development after extremely preterm birth is complex and poorly understood5 and the outcome in the very early preterm infants depends on a mixture of well-described haemorrhagic or ischemic injuries together with less known maturational and trophic disturbances affecting the developing brain.6 A range of perinatal factors, such as mother’s diseases and habits – e.g. nutrition,7 alcohol abuse,8 etc. – infection and inflammation,3,9 fetal growth restriction10,11 as well as the hospital course and exposure to systemic illness – chronic lung disease, postnatal infections12-15 – have been shown to further exacerbate disturbances in brain maturation with a negative impact on neurodevelopmental outcome.
Recently, the EPICure study reported short term outcomes after extreme preterm birth (≤26 weeks gestation) in England, comparing two birth cohorts in 1995 and 2006: a higher survival rate has been observed.4 Although an overall increase in severe neurological impairment was shown, a bigger proportion of babies born at 24 and 25 weeks’ gestation survived.2
Aim of this retrospective study was to investigate the neurodevelopmental outcome and related perinatal risk factors in a population of extremely preterm babies (23+0 and 25+6 weeks of gestation) admitted to a single neonatal tertiary care unit over an 8-year period.
Materials and Methods
We retrospectively collected data on all extremely preterm babies with GA at birth between 23+0 to 25+6 admitted to the neonatal tertiary care center (intensive care unit) Department of Clinical Sciences and Community Health, Fondazione IRCCS Ca’ Granda, Università degli Studi di Milano, Milan, Italy, over the course of 8 years (since January 2003 till December 2010).
Prenatal and perinatal characteristics were recorded: spontaneous or assisted reproduction, multiple pregnancy, mode of delivery, prenatal steroids, obstetric complications as gestational hypertension, severe intrauterine growth restriction (IUGR), preterm premature rupture of membranes (PPROM), maternal fever >38°C.
The following neonatal variables and main complications during the neonatal period were collected: gender, GA based on the last menstrual period and first-trimester ultrasonogram, birth weight, singletons or twins, Apgar score at 5 minutes, pneumothorax, patent ductus arteriosus (PDA) requiring any treatment, bronchopulmonary dysplasia (BPD) defined as oxygen dependency up to 36 weeks corrected age, rethinopathy of prematurity (ROP) requiring any treatment, abdominal surgery, any congenital malformation.
Infants with birth weight >10th or ≤10th percentile for GA, according to the North-Italian growth charts, were classified as appropriate for gestational age (AGA) or small for gestational age (SGA) respectively.16 All babies underwent brain magnetic resonance imaging (MRI) at term corrected age (TCA). MR images were classified as normal, with minor abnormalities and severely abnormal (cystic periventricular leukomalacia, arterial or venous stroke, multiple white matter punctate lesions, post-hemorrhagic ventricular dilatation). Infants were followed-up by assessments of neurodevelopment at two years of corrected age. The neurofunctional scale (NFS)17 and the Griffith’s Developmental Scales II (GMDS II)18 were used. The neurofunctional scale was assigned as follows: 0, normal function; 1, mild impairment of function (no limitations); 2, moderate impairment of function (possible but limited); 3, severe impairment of function (possible only with the use of facilitators or assisted devices); 4, function not possible. The Griffith’s scales consist of six sub-scales (locomotor, personal-social, hearing–speech, eye-hand coordination, performance and logical scales) which are combined to give the total score, the general quotient (GQ). The Griffith’s GQ was defined as: normal ≥88, mild cognitive delay between 76 and 87, severe cognitive deficit <76. Severe disability was defined as either NFS score ≥3 or GQ<76.
Analysis of variance and multivariate analysis were performed to evaluate the effect of tested factors on outcomes.
One hundred twenty-two infants were born during the study period; 55 babies (45%) survived to discharge but 9 were lost at follow-up. Twenty-six out of the 46 babies followed-up (56%) were born at 25 weeks’ gestation, 16 (35%) at 24 and 4 (9%) at 23.
The survival rates according to gestational age are shown in Table 1: survival rate doubled for each additional gestational week from 23 to 25 but no significant differences were observed between two study periods (period 1: 2003-2006; period 2: 2007-2010). Prenatal, perinatal and neonatal variables are shown in Tables 2 and 3.
Forty-six infants were evaluated by both neurofunctional assessment and Griffith’s scale at 2 years CA; results are reported in Figure 1. The neurodevelopmental outcome clearly differed according to gestational age at birth: none of the children born at 23 weeks had a normal neurofunctional or neurobehavioral assessment.
At 2 years of age 34/46 babies (74%) had a NFS score <3, without any significant change between the 2 study periods (77% in period 1 versus 72% in period 2). Ten to sixteen (62%) babies born at 24 weeks and 24/26 (92%) born at 25 weeks had NFS score <3. A Griffith’s GQ≥76 was observed in 26/46 (56%) babies: 7/16 (44%) born at 24 and 19/26 (73%) born at 25 weeks.
At the analysis of variance PPROM, absence of prenatal steroids, IUGR, male, lower GA, and severely abnormal MRI were significantly associated with higher NFS score (Table 4) and lower mean Griffith’s GQ (Table 5) at 2 years CA.
At the multivariate analysis gestational age, gender and the presence of major brain abnormalities remained significantly associated with severely impaired NFS while no independent factor was identified for the Griffith’s GQ (Tables 6 and 7).
Our results strongly support the evidence that survival as well as long-term neurobehavioral outcomes in extremely premature infants are significantly related to gestational age at birth: the lower the gestational age, the higher the risk for death or adverse outcome.
The survival rates observed in our population of infants are similar to those reported by larger studies as the EPICure study4 and almost double for each additional gestational week from 23 to 25 weeks. Despite the advances in perinatal care, we did not observe any significant improvement over an 8-year period, when two different study periods were considered.
Neurobehavioral development appears to be mainly related to gestational age with the most unfavorable outcome occuring in babies born at 23 weeks’ gestation, at the limit of viability. Griffith’s GQ was ≥76 in 62% and ≥88 in 33% of babies born between 24 and 25 weeks. No severe motor disabilities were found in 81% of babies born between 24 and 25 weeks.
The poor outcome of babies born at 23 weeks of gestational age was probably related to the small number of babies, all with unfavorable conditions at birth (male gender and inadequate prenatal steroid prophylaxis). Caution is needed when comparing results from different studies on long-term outcome in survivors of extremely preterm birth as differences in the neonatal management at birth (proactive approach) profoundly affect survival and rate of disability.
The occurrence of focal brain lesions plays a pivotal role in determining child neurodevelopment. Advances in MRI techniques have allowed an accurate assessment of brain development and gestational age is confirmed to be the main determinant of brain vulnerability to different types of brain lesions. Brain vulnerability is therefore age-dependent and the concept of gestationally determined brain vulnerability has been recently emphasized: the site and nature of the injury sustained being determined by a combination of the characteristics of the insult, the specific tissue and cell vulnerability and the gestation of the infant at birth.19
However, even in the absence of overt brain lesions, several perinatal and postnatal factors have been demonstrated to affect the microstructure of the developing white and grey matter with a potential impact on long-term neurodevelopmental outcome.
In our study IUGR, unlike SGA, was associated with an increased risk for adverse neurobehavioral outcome although it was not significant at the multivariate analysis. Intrauterine growth restriction, due to placental insufficiency leading to reductions in placental nutrient transfer, hypoxia, and progressive metabolic deterioration in almost any fetal organ system, has been associated with increased risk of perinatal mortality and morbidity.20-23 Several follow-up studies have described neurological impairments and worse neurodevelopmental outcomes in children born with IUGR.21,24 However, long-term neurological impairment frequently seen in IUGR infants cannot be attributed to the presence of overt brain lesions. The risk of brain damage in IUGR neonates could be related to both intrauterine compromised and premature birth. We recently demonstrated that, although gestational age at birth is the predominant factor affecting cerebral maturation, IUGR babies with brain sparing in utero have a mild delay in brain myelination. We speculated that in our study the lack of statistical association between IUGR and worse outcome at the multivariate analysis might be related to the small number of babies.11
PPROM has been reported to be risk factor for unfavorable outcome, probably due to the underlying physiopathological mechanisms related to chorioamnionitis;25,26 however, in this study we had no data on placental histology.