Neonatal strokes are cerebrovascular diseases resulting in injury to the developing brain between 20 weeks gestation and 28 days postnatally. Arterial ischemic strokes (AIS) are the most common stroke subtype during this period of rapid brain development. To date, there are no longitudinal studies that characterize neurocognitive trajectories and predictors in neonatal AIS. Current research has been cross-sectional and yielded contradictory results, with some studies demonstrating age-appropriate functioning post-neonatal AIS and others demonstrating neurocognitive deficits (11-41%). This discrepancy is attributable to methodological heterogeneity including small cohorts, varying age at assessment, differing/non-standardized measures, limited follow-up, and differing stroke subtypes. In this study, we explored neurocognitive trajectories, including predictors of heterogeneity, across infancy and early childhood post-neonatal AIS.

This was a retrospective cohort study of children having experienced neonatal AIS with longitudinal follow-up at The Hospital for Sick Children (N=40). Children underwent developmental assessment(s) through the Neonatal Neurodevelopmental Follow-up Clinic at 18- and/or 36-months (Bayley) and neuropsychological assessment(s) through the Children’s Stroke Program from age 4-13 years (WPPSI/WISC), with neurocognitive scores superimposed across comparable domains. Sex, lesion volume, lesion laterality, neurological diagnoses, and medical comorbidities were examined as potential predictors. Exploratory multilevel growth curve modelling was used to assess longitudinal neurocognitive trajectories, and to examine whether predictors had main effects or moderated the rate of change. Age (years) was the measure of time.

Of the 40 participants (60% male; 7.5% premature; 15% congenital heart disease; 2.5% genetic condition), 5 completed two (M_age1=1.60, IQR:1.52-1.71; M_age2=3.07, IQR:2.99-3.29), 21 completed three (M_age3=5.18, IQR:4.44-5.77), 11 completed four (M_age4=6.95, IQR:6.08-8.22), and 3 completed five (M_age5=10.25, IQR:9.44-10.66) assessments. Relative to the unconditional means model, the unconditional linear growth model had improved model fit when including time as a predictor of change (χ²=22.35, df=3, p<0.001). Baseline neurocognition was extrapolated at stroke occurrence and was in the average clinical range (102.48). A negative effect of time indicated longitudinal neurocognitive decline (β=-1.45, 95% CI:-2.51,-0.40, p=0.008). Individual differences in the rate of change indicated diverse neurocognitive trajectories (SD=1.81, 95% CI:0.77,4.28). Of the variables explored within conditional models, only lesion volume moderated the rate of change, with larger lesion volume associated with greater neurocognitive decline (β=-2.03, 95% CI:-3.56,-0.51, p=0.01); model fit was improved relative to the unconditional growth model (χ²=7.03, df=6, p=0.03). In examining potential main effects of the other variables, medical comorbidities contributed to lower baseline neurocognition (β=-11.97, 95% CI:-23.27,-0.68, p=0.04), above and beyond the moderating role of lesion volume; model fit was improved relative to the initial conditional model (χ²=4.36, df=8, p=0.04). Evaluation of sex, lesion laterality, and neurological diagnoses yielded non-significant results.

This is the first study to examine longitudinal neurocognitive trajectories in neonatal AIS, with the decline observed across infancy and early childhood in keeping with an early vulnerability hypothesis. Medical comorbidities, including congenital heart disease and genetic conditions, contributed to lower neurocognition at stroke occurrence. Larger lesion volume was associated with greater neurocognitive decline. Improved understanding of predictors of neurocognitive trajectories will inform the early identification of high-risk groups and the development of precision-based interventions.