Study design

This is an exposure-based cohort study between January 2023 and December 2027.

Study setting

Participants are recruited from the women’s and children’s services of the Royal Infirmary of Edinburgh (RIE), NHS (National Health Service) Lothian. The RIE provides maternity and newborn services for residents of the City of Edinburgh and the Lothians. It receives 7000 deliveries annually and is the regional centre for all neonatal intensive care in South East Scotland. Approximately 100 infants with a birth weight of <1500 g receive intensive care at the RIE per annum.

Study population

We plan to recruit 300 mother–infant dyads: 200 preterm deliveries with GA <32 weeks (exposed cases) and 100 term deliveries with GA >37 weeks (non-exposed comparators). GA is determined by the first trimester ultrasound scan. Preterm infants are included if a mother booked her pregnancy and delivered at the RIE (study centre) or if a mother booked her pregnancy at a hospital outside the study centre but was transferred to it with her baby in utero due to planned or expected birth <32 weeks.

Exclusion criteria: (1) Preterm infants who are transferred to the study centre postnatally for intensive care; (2) Infants with congenital anomalies: structural or functional anomalies (eg, metabolic disorders) that occur during intrauterine life and can be identified prenatally, at birth or later in life (WHO definition) and (3) Infants with a contraindication to MRI at 3Tesla determined by the Edinburgh Imaging safety policy, which is developed in accordance with UK Medicines and Healthcare Products Regulatory Agency safety guidelines.

Participant selection and enrolment

Women who present to the RIE with threatened preterm labour and for whom delivery is planned or expected at less than 32 weeks GA. The comparator group (non-exposed term infants) are born to women who attend the RIE for antenatal care or delivery at >37 weeks GA. Potential participants are identified using NHS systems: maternity TRAK and the neonatal electronic patient record. As with prior work,37 this will result in a sampling distribution with fewer GA values between 32 and 37 weeks but will maximise sampling at important ends of the distribution within practical funding and recruitment constraints. Our analytical strategy, outlined below, will therefore benefit from a relative increase in power under consideration of important assumptions which apply for some but not all variables, including that a linear dose-response effect is present across the GA continuum between term and preterm.38 39

Consent to enter the study is sought from each participant after a full explanation has been given, an information leaflet offered and time allowed for consideration. Signed participant consent is obtained in two stages for the preterm group: first, for data collection from the antenatal period to the first week of postnatal life, and second, for data and samples over the rest of the neonatal period to the end of the study. Signed participant consent for all aspects of the study will be obtained in one stage for the comparator group. Consent to recontact for follow-on studies subject to additional funding is sought.

Outcomes

Identification of targets in neuroendocrine stress and immune pathways that lead to atypical brain development in preterm infants indexed using 3 MRI markers of EoP: morphometric similarity networks (MSNs),40 hierarchical complexity (HC)41 and magnetisation transfer saturation imaging (MTsat).42 Table 1 summarises the assessment schedule, data collection methods, sample type/domain and the test or task. Data from cases and comparators are collected using the same data collection instruments.

Questionnaire and records

Demographic and clinical information is extracted from the maternal and infant records. The tools to assess cognition, behaviour, well-being and family circumstances are listed in table 1.

Neuroimaging

Participants are scanned using a Siemens MAGNETOM Prisma 3T MRI clinical scanner (Siemens Healthcare, Erlangen, Germany). For those at term-equivalent age, a 16-channel phased-array paediatric head receive coil is used to acquire sagittal three-dimensional (3D) T2-weighted (T2w) sampling perfection with application-optimised contrasts by using flip angle evolution (SPACE; 1 mm isotropic resolution, echo time (TE)=409 ms, repetition time (TR)=3200 ms), axial spin-echo echo-planar imaging multishell diffusion MRI (dMRI; 2 mm isotropic; 3×b=0 with reverse phase encoding, 16×b=0, 3×b=200, 6×b=500, 64×b=750, 64×b=2500 s/mm² with optimal angular coverage43; TR/TE=3500/78 ms), sagittal 3D T1-weighted (T1w) magnetisation-prepared rapid acquisition with gradient echo (MPRAGE; 1 mm isotropic, TR/TE=1970/4.69 ms, inversion time (TI)=1100 ms, flip angle (FA)=9°) and B₁⁺ field mapping (2.59×2.59×3.00 mm) scans. Magnetisation transfer (MT) saturation (MTSat) imaging is acquired, comprising three sagittal multiecho spoiled gradient echo scans (1.6 mm isotropic, TE=2.21, 6.31, 10.41 ms): (1) with gaussian MT preparation pulse (offset 1200 Hz, duration 9.984 ms, FA=500°; TR=75 ms, FA=5°); (2) proton-density weighted (PDw; TR=75 ms, FA=5°) and (3) T1w (TR=15 ms, FA=14°); an additional 8 echoes are acquired during the PDw scan to facilitate T₂* and quantitative susceptibility mapping (TE=15.00, 20.00, 25.00, 30.00, 35.00, 40.00, 45.00, 50.00 ms).

If the infant stays settled, axial 3D susceptibility-weighted (0.75×0.75×3.0 mm, TR/TE=28/20 ms) and axial 2D fluid-attenuated inversion-recovery (FLAIR) BLADE (0.94×0.94×3.0 mm, TR/TE/TI=10 000/130/2606 ms) scans are acquired. Tissue heating and acoustic noise exposure are limited through active noise cancellation and by appropriately setting the gradient slew rate and other pulse sequence parameters. Participants are scanned in normal mode with respect to tissue heating and peripheral nerve stimulation. Further details of the protocol are provided in online supplemental file 1.

Conventional images are reported by a paediatric radiologist using a structured system.44 45 We use established methods to derive three markers of EoP: MSNs,40 HC41 and MTsat.42 Images are processed to derive features for secondary analyses, including but not limited to tract segmentations46 47 and structural regions of interest.48

HPA axis activity (umbilical cord blood and maternal and neonatal hair)

Laboratory analyses of corticosteroids and their precursors and metabolites in plasma (2 mL) and hair (>0.3 cm 2 cm from neonates, up to 3 cm from mothers) are conducted at the University of Edinburgh Clinical Research Facility Mass Spectrometry Core. We have developed a robust method for steroid extraction from plasma (100 µL) and tissues,49 with quantification of cortisol and related corticosteroids, including cortisone, as well as dexamethasone and its metabolites, simultaneously by liquid chromatography tandem mass spectrometry, using a Sciex QTRAP 6500 (Warrington, UK) operated in positive ion electrospray ionisation with a Waters Acquity UPLC system (Manchester, UK).50

DNAm (saliva)

DNA from saliva is extracted using prepIT.L2P reagent (DNA Genotek, Ontario, Canada). DNA will be bisulfite converted and methylation measured using Illumina HumanMethylationEPIC BeadChip (Illumina, San Diego, California, USA) at the Edinburgh Clinical Research Facility Genetics Core, Edinburgh, UK. The epigenetic measures of immune function, EpiScores, are calculated for each participant.29

Dried blood spots (umbilical cord blood and neonatal dried blood spot)

Blood spots will be collected using Schleicher and Schuell 903 filter paper (6×3.2 mm spots per subject). Cards are stored at −20°C in the Centre for Reproductive Health and analysed in batch, subject to funding.21 22 37

Placenta

Samples are stored at −80°C in the Edinburgh Reproductive Tissue BioBank for future analyses subject to approvals.

Gut microbiome (faeces)

The gut microbiome plays a role in human health and disease, including child development,51 and is modified by age at birth, sex, mode of delivery, antibiotic exposure and feed type.52–55 The microbiome may mediate interactions of the preterm gut–brain axis.56–58 Three faecal samples are collected from cases during NICU care, and one from comparators within 2 weeks of birth. Maternal faecal samples will be collected. Samples will be processed and stored at −80°C for later analyses, subject to funding.

Outcomes measurement

Samples and data will be collected at three time points in the perinatal period (table 1).

Data analysis

Image processing is carried out at the University of Edinburgh using established pipelines for MSNs,40 HC41 and MTsat.42 There are two statistical approaches. In the first, multivariable regressions in a predictive framework are used to test whether PTB-RFs are associated with MSNs (brain age), HC (connectome architecture) and MTsat (a marker of myelination). For this, we will use standard statistical approaches (hypothesis testing, statistical association computations) and also machine learning methods ranging from feature selection to statistical mapping using widely used tools such as random forests and support vector machines.59–61 In the second, mediation analyses within a structural equation modelling framework are used to investigate the role of neuroendocrine stress activation and chronic inflammation as mediators of PTB-RFs on features of EoP.29 This simultaneously characterises associations among HPA axis activation/DNAm, PTB-RFs, and brain features and specifically tests the hypothesis that stress and/or chronic inflammation partly and significantly mediate associations between PTB-RFs and brain development.

PTB-risk factor selection

PTB-RF selection is informed by the results of a national population-based cohort study that is a part of the PRENCOG programme. In summary, the weighted contributions of multidimensional PTB-RFs to neurodevelopmental outcomes and the real-world educational performance of children born preterm will be determined by linking the electronic health records of >100 000 infants born in England and held in the National Neonatal Research Database (NNRD) to the National Pupil Database (NPD). The NNRD is a Health Research Authority-approved National Information Asset that contains detailed, quality-assured data (Neonatal Data Set; NHS Information Standard DAPB1595) extracted from Electronic Patient Records.62 63 The NPD is a key Department for Education data store covering attainment for learners in England. Ethical and regulatory approvals for this record linkage and analysis are granted to CB64 (REC reference 21/EM/0130).

Sample size

The sample size for groupwise comparisons of image data using biological variables is based on properties of the chosen EoP image phenotypes,40 41 and term and preterm differences we have observed in predictor variables (1) hair cortisol concentrations15: 401 pg/mg (262–615) vs 82 pg/mg (55–169), respectively and (2) group differences in DNAmCRP EpiScores.26

To maximise reproducibility, we will use (1) open-access neuroimaging protocols and standard operating procedures (SOPs) for sampling and analysis of biosamples, (2) behavioural assessment with clinical, dimensional and trait measures, multiple informants, direct observation and biometric data, (3) recommended reporting standards for neuroimaging, HPA axis activity and DNAm, (4) prespecified blind data processing, (5) analysis preregistration and (6) source code and data sharing. All manuscripts will be posted on preprint sites to facilitate another layer of peer review including critical insights into methodology.

Patient and public involvement

The research questions were informed by parent priorities for research about childhood outcomes following PTB,65 attitudes of longitudinal cohort participants towards recent opportunities and controversies within health data science66 and stakeholders. The stakeholder groups are the Adult Preemie Advocacy Network (APAN, a network of adults who were born preterm, coauthor LI) and an eight-member parent advisory group. Stakeholders codesigned the research questions, reviewed the content of all participant-facing materials, including the participant information sheet and graphics (figures 2 and 3), and informed our dissemination strategy. We commissioned a graphical design artist to create the PRENCOG study logo, a participant-facing infographic and a video animation to support recruitment (https://media.ed.ac.uk/media/Prencog_Neonatal/1_9llqdgsd).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Codesigned PRENCOG logo, available in black and white.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Codesigned participant-facing infographic.

Related work

Children born preterm and term comparators enrolled in a separate longitudinal study (Theirworld Edinburgh Birth Cohort, TEBC) are invited for behavioural assessments at age five, as described in the TEBC protocol.37 With funding from the PRENCOG programme, MRI data are acquired at this time point. We have added the following behavioural tasks to those listed in the TEBC protocol: Theory of Mind booklet task,67 executive functions (Early Childhood Inhibitory Touchscreen Task,68 CORSI block tapping task69 and prohibited toy task),70 exploratory play (Novel toy task) and reading (from the Woodcock-Johnson IV subscales).

The goal is to define the functional and structural neural substrates of critical cognitive functions in preterm children and, using perinatal data, characterise factors that shape neurocognitive development at 5 years of age.

Five-year-old participants are scanned using a 32-channel phased-array adult head receive coil to acquire sagittal 3D T1w MPRAGE (1 mm isotropic, TR/TE/TI=2500/4.69/1180 ms, FA=7°), sagittal 3D T2w SPACE (0.9 mm isotropic, TR/TE=3200/407 ms), axial 2D T2w FLAIR (0.94×0.94×3.0 mm, TR/TE/TI=9500/124/2556 ms) and axial spin-echo echo-planar imaging multishell dMRI (2 mm isotropic; 3×b=0 with reverse phase encoding, 15×b=0, 3×b=200, 6×b=500, 64×b=1000, 64×b=2000 s/mm² with optimal angular coverage43; TR/TE=2800/82 ms) scans. MTsat imaging is acquired, comprising three sagittal multiecho spoiled gradient echo scans (1.6 mm isotropic, TE=2.29, 6.33, 10.37 ms): (1) with MT preparation pulse as above (TR=35 ms, FA=5°), (2) PDw (TR=35 ms, FA=5°) and (3) T1w (TR=15 ms, FA=18°). B₀ field mapping (2.3 mm isotropic) is acquired prior to three functional MRI scans, which are acquired using 2D gradient echo echo-planar imaging (2.3 mm isotropic, TR/TE=1000/30 ms, FA=60°). During the functional MRI scans, children view selected movies that are age appropriate, engaging and enable characterising neural correlates of several cognitive functions. Further details of the protocol are provided in online supplemental file 2.

We use an information booklet, an animation (https://media.ed.ac.uk/media/PRENCOG_5YEAROLDAPPOINTMENT_ANIMATION/1_akzmmsc4) and a mock scan to acclimate 5-year-old participants to the MRI environment and to train them to stay very still (ie, <2 mm motion).71 The children use in-ear headphones to listen to the soundtrack of movies and for communication with the researchers operating the scan in the control room. The researchers communicate with children approximately every 5 min during the scan; children respond by speaking aloud. The in-ear headphones reduce the MRI noise to safe levels; soft pads offer additional hearing protection and help to stabilise children’s heads. An additional member of the research team stands near the child’s feet and the bore of the scanner to monitor the child during the scan. If the child moves, this researcher pats the child’s leg as a reminder to stay still.

We will use functional, structural and diffusion MRI data to investigate differences in brain structure and function as a function of GA, in 5-year-old children. Our analyses will include focused studies of responses in specific functional networks that underly particular cognitive domains (eg, social cognition, attention, language, reading), as well as whole-brain studies characterising distributed impacts of PTB (eg, on white matter tract integrity, MTsat, network architecture and cortical morphology). These data also enable longitudinal studies of brain structure, given that participants completed structural and diffusion MRI scans as neonates. This line of work will build directly on evidence from the neonatal scans by testing for sustained impacts of GA on brain development at age 5 years.40–42 As described above, we will also investigate the relative roles of other risk and protective factors (eg, SES,48 maternal stress,12 infant nutrition72 and early linguistic environment73) on neurocognitive development in children born preterm.

Ethics and dissemination

Ethical favourable opinion for all neonatal studies has been obtained from the South East Scotland Research Ethics Committee 02 (23/SS/0067) and NHS Lothian Research and Development (2023/0150). A favourable ethical opinion for data collection and analyses in related work on 5 years has been provided by the South East Scotland Research Ethics Committee 01 (16/SS/0154).

Statistical analysis plans for the main analyses will be published on Open Science Framework.

Results will be reported to the UKRI Medical Research Council. They will be presented at national and international scientific conferences and summarised on a study-specific website in lay form and via a newsletter for families (https://www.ed.ac.uk/centre-reproductive-health/prencog). They will be published on preprint servers and in peer-reviewed publications. At the end of the programme of work, we will cocreate with stakeholders a scientific animation to illustrate the research insights and offer accessible and digestible information to families. Stakeholders (APAN) will disseminate the main findings via their social media channels and website. We will engage with the University of Edinburgh public relations and media office to ensure maximum publicity and benefit.

Safety considerations

We do not anticipate risk from any of the biosample collections or questionnaires.

The MRI scanner generates loud acoustic noise, so flexible earplugs and earmuffs are used to prevent noise discomfort and to encourage infants to sleep. We use established procedures described ensuring infant safety and physiological stability during imaging.37 The infant has continuous monitoring of vital signs (heart rate and oxygen saturation) with an MR conditional patient monitor. The attending clinical practitioner will record observations every 5 min until 1 hour after the infant has woken up, and the scan will be stopped if there are any abnormalities in monitoring. Full neonatal resuscitation facilities are available on site. SOPs for ensuring safety in the MRI environment are in place at the Edinburgh Imaging facility.