Longitudinal assessment of the impact of preterm birth on cognitive functions and identification of novel methods for stratification of children at risk of long-term impairment

Abstract

During the last few decades advances in perinatal care have led to significant improvements in the survival of preterm infants. However, survivors of preterm birth are at increased risk of brain maldevelopment, and subsequent neurodevelopmental deficits compared to their term-born peers. Given that neurodevelopmental trajectories are amenable to early interventions, there is a critical window of opportunity to improve long-term outcomes for children born preterm. Currently, there are no clinical tools which accurately predict brain growth or subsequent neurodevelopmental outcomes for preterm infants.

Hence, the objectives of this study were: (a) to identify the early life exposures that impact total and regional brain volumes at term-equivalent age following preterm birth, and (b) to accurately predict language outcomes at two years Corrected Gestational Age (CGA). To this end, we analysed data from a longitudinal cohort of preterm infants born before 33 weeks of gestation at the Royal Infirmary of Edinburgh and developed parsimonious machine learning models using feature selection and advanced machine learning techniques.

This work revealed that a combination of clinical, biological, and environmental exposures, including potentially modifiable risk factors, such as postnatal nutrition, respiratory illness, and socioeconomic deprivation, best predicts cerebral tissue volumes at term-equivalent age. Furthermore, we found that a combination of diffusion tensor imaging features and clinical perinatal factors collected as part of routine care accurately predict language outcomes at two years CGA. These results have important implications for clinical practice: mitigating these risk factors can inform current perinatal practices, leading to enhanced brain development and neurodevelopmental outcomes following preterm birth. Moreover, these findings could facilitate timely identification of infants who are at considerable risk of language impairment and who may benefit from targeted early interventions and support services. Overall, our research can potentially offer preterm infants a healthier start in life, improved long-term outcomes and a better quality of life.