Methods
Following a PECO framework, a systematic literature search was conducted in PubMed, Scopus, and CINAHL: population, infants born before 37 completed weeks of gestation; exposure, any enteral or parenteral nutritional exposure, supplementation strategy, fortification approach, or individualized nutritional intervention; comparator, an alternative nutritional strategy, lower or absent exposure, standard care or, when relevant, a term-born reference group; outcomes, MRI- or CUS-based measures of brain growth, maturation, connectivity, or injury. Studies reporting volumetric, morphometric, diffusion, connectivity, spectroscopy, or quantitative ultrasound outcomes were eligible. The methodology was adapted from the PROSPERO-registered review by Ottolini et al. Original studies published between 2020 and 2026 (1 January 2020-4 September 2026) evaluating nutritional exposures or interventions in preterm infants with magnetic resonance imaging (MRI) and/or cranial ultrasound (CUS) outcomes were included. Risk of bias was assessed using the Newcastle-Ottawa Scale (NOS), ROBINS-I and RoB 2 according to study design. Quantitative imaging studies were screened for eligibility for ATLAS assessment. Given the substantial heterogeneity in nutritional exposures, imaging protocols, timing of assessments, and neurodevelopmental outcomes, meta-analysis was not performed. Certainty of evidence was assessed at the level of each causal question and neuroimaging outcome domain using the GRADE framework. Randomized intervention evidence and observational exposure evidence were evaluated according to the design and target contrast contributing to each body of evidence. Judgments considered risk of bias, inconsistency, indirectness, imprecision, publication bias, and, where applicable, factors that could increase certainty. Statistical significance was not used to determine certainty, and studies arising from shared or partially overlapping parent populations were not counted as independent replication.
