Brain activity as a candidate biomarker for personalised caffeine treatment in premature neonates

Background Caffeine is one of the most frequently administered medicines in neonatology—prescribed for the management of apnoea of prematurity, to aid extubation and increasingly for conditions such as bronchopulmonary dysplasia. Caffeine guidelines for the management of apnoea of prematurity indicate use based on the age of the infant, but this does not account for individual variation in apnoea rate. Consequently, infants may risk caffeine undertreatment or adverse events due to over-exposure. Apnoea in preterm infants is related to nervous system immaturity, hence, as an essential first step to assess whether brain activity may be a useful biomarker for caffeine treatment, we tested the hypothesis that apnoea rate is related to brain activity. Methods In this single-centre prospective observational cohort study, we simultaneously recorded brain activity using electroencephalography (EEG) and respiration using impedance pneumography in 74 infants aged 31–36 weeks postmenstrual age (PMA) on 138 separate occasions. The primary outcome was the association between apnoea rate and brain age gap (defined as the difference between the infant's brain age and their PMA; brain age is calculated from brain activity using a deep learning algorithm) . In an exploratory sub-study, we compared the apnoea and desaturation rate in the 7 days after infants stopped caffeine treatment, between those infants with immature and mature brain activity. Results We demonstrate that apnoea rate in moderate/late preterm infants is dependent on brain age gap (p:0.024; β 95% CI:−0.22 −0.41 to −0.03). In contrast, apnoea rate was not correlated with PMA (p:0.58; β 95% CI:−0.04 −0.16 to 0.09). In the exploratory sub-study, we find that when caffeine is discontinued, infants with immature brain activity have more frequent apnoeas and desaturations compared with those with more mature brain function. Conclusions These findings provide initial evidence to indicate that brain age is a candidate biomarker for personalised caffeine treatment in preterm infants.