Contribuição da ecografia cerebral transfontanelar para a avaliação do crescimento cerebral do recém-nascido pré-termo

Tese de doutoramento, Medicina (Pediatria), Universidade de Lisboa, Faculdade de Medicina, 2013

Preterm birth contributes greatly to infant mortality and to significant morbidity in survivors. Despite this, the progress made during the past few decades in neonatal intensive care translates into improved survival of extremely preterm infants and a good quality of life for most survivors during childhood and adulthood. Major challenges in daily practice for the neonatologist are the ability to predict longer-term outcome, to explain the complexities of these issues to parents and the targeting of appropriate rehabilitation and support for very preterm infants and their families. The greatest challenges in terms of predicting outcomes are not generally present when dealing with infants that have major lesions detected by cerebral ultrasound and possibly better defined by magnetic resonance imaging, but when counselling the parents of babies who reach the term-equivalent age with apparently “normal” brain scans, because we know that some of these infants have poor outcomes that are difficult to predict using current standards. The only brain imaging tool that is widely available to support counselling is cerebral ultrasound and there is no evidence to support the use of magnetic resonance imaging for all very preterm infants at term-equivalent age, even in high resource settings. Thus, the aim of the research project underlying this thesis is to evaluate less explored contributions of conventional cerebral ultrasound, which were either not previously described or are not used routinely, in order to maximize the information obtained for all very preterm infants. The evaluations of less well explored uses of cerebral ultrasound contained in this thesis are: I. A tridimensional model to estimate cerebral volume at term-equivalent age; II. The usefulness of cerebral ultrasound for the diagnosis of non-cystic white matter injury; III. The utility of cerebellar measurements for evaluating cerebellar growth in very preterm infants during the neonatal period; IV. The utility of cerebellar measurements for estimating gestational age in very low birth weight infants in whom gestational age is uncertain. I - Very preterm infants are at particular risk of neurodevelopmental impairments. This risk can be anticipated when major lesions are seen on cerebral ultrasound. However, most preterm infants do not have such lesions yet many have a relatively poor outcome. Our study aims to describe a tridimensional cerebral ultrasound model at term-equivalent age to help in identifying infants without cerebral ultrasound lesions but with smaller cerebral volume. We scanned a cohort of very preterm infants at term-equivalent age and term-born controls. Infants with major cerebral lesions were excluded. Measurements of intracranial diameters (biparietal, longitudinal, height), brain structures, ventricles and extracerebral space were made. A mathematical model was built to estimate from the cerebral ultrasound measurements the axial area and volumes of the cranium and brain. Clinical data were extracted from the medical records. Appropriate statistical methods were used for comparisons; p-value <0.05 was considered significant. We assessed 128 infants (72 preterms and 56 controls). The preterms’ head at term-equivalent age was longer (11.5 vs. 10.5 cm, p<0.001), narrower (7.8 vs. 8.4 cm, p<0.001) and taller (8.9 vs. 8.6 cm, p<0.01) than the controls and the extra-cerebral spaces and the ventricles were larger. Estimated intracranial volume was not statistically different between the groups (411 vs. 399 cm3, NS), but the preterms had larger estimated extracerebral space volume (70 vs. 22 cm3, p<0.001) and smaller estimated cerebral volume (340 vs. 377 cm3, p<0.001). On bivariate analysis, significant risk factors for a smaller brain volume were lower gestational age and birth weight, being small-for-gestational age, higher CRIB score, need for invasive ventilation, chronic lung disease, hypotension and necrotizing enterocolitis, but only the CRIB score remained significant in a multivariate model. Our data suggest that even in the absence of major cerebral lesions, the average extra-uterine cerebral growth of very preterm infants is compromised. Our model can help identify those preterm infants with smaller brains. Later follow-up data will determine the neurodevelopmental outcome of these preterm infants in relation to their estimated brain volumes. II - Periventricular white matter is particularly vulnerable in very preterm infants. Nowadays cystic periventricular leukomalacia is a rare entity, and non-cystic white matter injury, known as non-cystic periventricular leukomalacia, is the commonest “lesion” affecting the preterm brain. There is no consensus about whether non-cystic periventricular leukomalacia can be reliably identified from cUS or whether there is any reliable correlate of non-cystic periventricular leukomalacia on cerebral ultrasound at term-equivalent age. We aim to compare brain volumes and linear measures at term-equivalent age in infants with and without a diagnosis of non-cystic periventricular leukomalacia. A cohort of preterm infants under 32 weeks gestational age without major lesions were serially assessed using cerebral ultrasound. Non-cystic periventricular leukomalacia was defined as periventricular white matter echogenicity comparable to the choroid plexus seen on two scans at least two weeks apart after the first post-natal week; a diagnosis was achieved by consensus. At term-equivalent age, infants were scanned for the estimation of brain volume and for measuring ventricular dimensions, corpus callosum length/thickness and central grey matter width. Head circumference was also measured. The data were compared between those with/without non-cystic periventricular leukomalacia. Observer agreement was assessed using kappa statistic. Of 63 eligible infants 29% had non-cystic periventricular leukomalacia. Significant differences were found between those with/without non-cystic periventricular leukomalacia for 5 minute Apgar score, CRIB score, invasive ventilation rates and chronic lung disease but not other relevant clinical data. No significant differences were found for estimated brain volume, ventricular size, corpus callosum length/thickness or central grey matter width. Intra-observer reliability was moderate (kappa=0.51-0.56) and inter-observer reliability was poor (kappa=0.20-0.32). The results of this study indicate that cerebral ultrasound diagnosis of non-cystic periventricular leukomalacia as we defined it should not be used as a predictor of lower brain growth detectable at TEA, given that infants with this diagnosis showed identical estimated cerebral size at term-equivalent age when compared to babies with normal scans. Additionally, there was an unsatisfactory intra and inter-observer reliability, suggesting a degree of subjectivity in this cerebral ultrasound diagnosis despite the application of well-defined criteria. III - Reduced supra-tentorial brain growth has been shown in preterms by term-equivalent age, but cerebellar growth may be preserved in the absence of supra-tentorial injury. Our study aimed to compare cerebellar size assessed using cerebral ultrasound at term-equivalent age between preterm infants and term-born controls. Cerebellar dimensions (including transverse cerebellar diameter, vermis height, antero-posterior vermis diameter, cerebellar vermis area) were measured using ImageArena® software in 71 infants born at less than 32 weeks gestation and without significant scan abnormality at term-equivalent age and 58 term controls. Observer agreement was evaluated. Compared to controls, preterms at term-equivalent age had smaller transverse cerebellar diameter (4.9 vs. 5.2cm; p<0.001) but larger cerebellar vermis area (4.7 vs. 4.3cm2; p<0.005) and antero-posterior vermis diameter (2.4 vs. 2.2cm; p<0.001) but these differences were no longer seen after correcting for head shape. In very preterm infants born before 28 weeks gestation and in small-for-gestational age preterms the cerebellar vermis area measurements were statistically similar to controls, whereas transverse cerebellar diameter was still smaller. Our data support neonatal sparing of preterm cerebellar growth that is measureable using cerebral ultrasound, even in the most immature and small-for-gestational age infants. We suggest cerebral ultrasound can be used to assess cerebellar size at term-equivalent age and thus may be a useful tool for detecting infants with poorer cerebellar growth who are at increased risk of disability. IV - Clinical assessment of gestational age in newborn preterm infants can be challenging. Several cerebral ultrasound approaches for estimating gestational age using cerebellar measurements are reported, claiming to be simpler and more accurate than clinical assessment, but they are not widely used. We aimed to compare the accuracy of four previously described measurements and compare their use in preterm infants. We studied infants born at less than 32 weeks gestation defined by in-vitro fertilization date or early fetal ultrasound, excluding infants with neurological problems. Vermis anterior-posterior diameter, vermis height, and transverse cerebellar diameter via anterior and mastoid fontanelles were measured. Estimated post-menstrual age was calculated using published equations, and compared to known post-menstrual age using intraclass correlation coefficient. Intra and inter-observer reliability were determined. We studied 80 infants (mean gestation 28.5 weeks [range 24-32] and mean postnatal age 5.7 days). Intraclass correlation coefficient was 0.761 for vermis antero-posterior diameter, 0.632 for vermis height, 0.115 for transverse cerebellar diameter measured via the anterior fontanelle and 0.825 for transverse cerebellar diameter measured via the mastoid fontanelle. The equation for transverse cerebellar diameter measured via the mastoid fontanelle gave the best estimate of gestation (mean estimate -2 days; 95%CI±13.8 days). The actual measurements of transverse cerebellar diameter were similar whether they were made using the anterior and of mastoid fontanelles as was the accuracy of estimating gestational age using the equation for transverse cerebellar diameter measured via the mastoid fontanelle in both appropriately grown and small-for-gestational age infants. Inter and intra-observer reliability was very good for all measurements. We found that three of the four previously described equations for estimating gestational age from cerebellar measurements gave good estimates of gestational age in preterm infants. The equation described for the transverse cerebellar diameter measured via the mastoid fontanelle gave the narrowest 95% confidence interval. We recommend the transverse cerebellar diameter measured via the mastoid fontanelle equation for the estimation of gestational age in very low birthweight infants but the transverse cerebellar diameter measurement itself can be performed either via the anterior or mastoid fontanelle. The contribution of some of the less explored approaches to the use of cerebral ultrasound described in this thesis may enhance the identification of very preterm infants that are most at risk of neurodevelopmental problems despite having a “normal” scan during their neonatal period and at term-equivalent age. Since the widespread use of magnetic resonance imaging at term-equivalent age for all very preterm infants is not a reality in the coming years (and the interpretation of magnetic resonance imaging findings in preterms at term-equivalent age and their significance for later outcomes still needs to be defined), improving the information obtained from cerebral ultrasound is of great value in aiding clinicians with counselling parents about their child’s outcome and for selecting the most “at risk” infants for magnetic resonance imaging and/or more intensive follow-up and intervention.