Interleukin-8 dysregulation is implicated in brain dysmaturation following preterm birth

BACKGROUND
Preterm birth is associated with dysconnectivity of structural brain networks, impaired cognition and psychiatric disease. Systemic inflammation contributes to cerebral dysconnectivity, but the immune mediators driving this association are poorly understood. We analysed information from placenta, umbilical cord and neonatal blood, and brain MRI to determine which immune mediators link perinatal systemic inflammation with dysconnectivity of structural brain networks.


METHODS
Participants were 102 preterm infants (mean gestational age 29+1 weeks, range 23+3-32+0). Placental histopathology identified reaction patterns indicative of histologic chorioamnionitis (HCA), and a customized immunoassay of 24 inflammation-associated proteins selected to reflect the neonatal innate and adaptive immune response was performed from umbilical cord (n=55) and postnatal day 5 blood samples (n=71). Brain MRI scans were acquired at term-equivalent age (41+0 weeks [range 38+0 - 44+4 weeks]) and alterations in white matter connectivity were inferred from mean diffusivity and neurite density index across the white matter skeleton.


RESULTS
HCA was associated with elevated concentrations of C5a, C9, CRP, IL-1β, IL-6, IL-8 and MCP-1 in cord blood, and IL-8 concentration predicted HCA with an area under the receiver operator curve of 0.917 (95% CI 0.841 - 0.993, p<0.001). Fourteen analytes explained 66% of the variance in the postnatal profile (BDNF, C3, C5a, C9, CRP, IL-1β, IL-6, IL-8, IL-18, MCP-1, MIP-1β, MMP-9, RANTES and TNF-α). Of these, IL-8 was associated with altered neurite density index across the white matter skeleton after adjustment for gestational age at birth and at scan (β = 0.221, p=0.037).


CONCLUSIONS
These findings suggest that IL-8 dysregulation has a role in linking perinatal systemic inflammation and atypical white matter development in preterm infants.


Introduction
Preterm birth affects around 15 million births annually (Chawanpaiboon et al., 2019), and is an important cause of cerebral palsy, cognitive impairment, autism spectrum disorder and psychiatric disease later in life (Twilhaar et al., 2018). Magnetic resonance imaging (MRI) studies reveal a cerebral phenotype of preterm birth that is associated with later function, and includes diffuse white matter disease, T prematurity, which can be conceptualized as white matter injury and subsequent dysmaturation of diverse cellular processes resulting in atypical development of white and grey matter structures (Hagberg et al., 2015;Volpe, 2019). Preterm infants have a distinct inflammatory profile in blood and cerebrospinal fluid (CSF) that includes higher levels of pro-inflammatory cytokines and lower levels of neuroprotective growth factors compared to infants born at term (Boardman et al., 2018;Skogstrand et al., 2008), and there is evidence that both systemic inflammation and neuroinflammation are associated with overt forms of preterm brain injury. For example, elevated IL-1β, IL-6, IL-8 and IL-10 in umbilical cord and early postnatal blood is associated with intraventricular hemorrhage and white matter lesions soon after birth (Duggan et al., 2001;Leviton et al., 2018;Yoon et al., 1996), and persistently elevated pro-inflammatory proteins during the first 2 weeks after preterm birth are associated with increased risk of cerebral palsy (Carlo et al., 2011;Kuban et al., 2014) and impaired neurocognitive development in infancy and childhood (Hansen-Pupp et al., 2008;Kuban et al., 2017;Leviton et al., 2016;O'Shea et al., 2012O'Shea et al., , 2013Yanni et al., 2017). Furthermore, specific co-morbidities of preterm birth characterized by systemic inflammation, including intrauterine inflammation (histologic chorioamnionitis, HCA) and necrotizing enterocolitis, are associated with abnormal white matter on magnetic resonance imaging (Anblagan et al., 2016;Ball et al., 2017;Barnett et al., 2018;Shah et al., 2008). However, previous study designs leave uncertainty about which immune mediators are associated with white matter dysmaturation and network dysconnectivity.
Diffusion tensor MRI (DTI) and neurite orientation dispersion and density imaging (NODDI) support inference about the microstructural properties of developing white matter (Pietsch et al., 2019;Tariq et al., 2016;Zhang et al., 2012). Specifically, normal brain maturation is characterized by a reduction in white matter mean diffusivity (MD) and an increase in neurite density index (NDI); but MD is increased and NDI decreased in preterm infants at term equivalent age, compared with healthy controls infants born at term (Pecheva et al., 2018). These changes reflect an increase in water content and a decrease in white matter organization in preterm infants. Peak width of skeletonised mean diffusivity (PSMD) is a method for histogram based calculation of MD distribution across the entire white matter skeleton, which provides a measure of generalized white matter microstructure and dysconnectivity that is related to cognition (Baykara et al., 2016;Deary et al., 2019;Wei et al., 2019). In previous work we extended the histogram model to include peak width of skeletonised neurite density index (PSNDI) and found that PSMD and PSNDI are both altered in preterm infants at term equivalent age, indicating that these are useful biomarkers of generalized white matter connectivity in the developing brain .
Identification of immune mediators of generalized dysconnectivity during the perinatal period is important for elucidating new targets for neuroprotection therapies. In this study, we utilized an immunoassay of 24 analytes customized to reflect perinatal innate and adaptive immune responses, and analyzed profiles from umbilical cord and postnatal blood samples with placental histopathology and brain MRI to test the hypothesis that specific immune mediators link systemic inflammation with atypical white matter development in preterm infants. We found that elevated IL-8 is a characteristic feature of intrauterine and postnatal systemic inflammation, and it is associated with white matter dysconnectivity in preterm infants.

Participants
Participants were 102 preterm infants born ≤32 +0 weeks' gestation, delivered at the Royal Infirmary of Edinburgh, UK and recruited to a longitudinal study of the effect of preterm birth on brain development and long term outcome . Exclusion criteria included infants with major congenital abnormality, post-hemorrhagic ventricular dilatation, cystic periventricular leukomalacia, or contraindications to MRI. Ethical approval was obtained from the UK National Research Ethics Service and parents provided written informed consent (South East Scotland Research Ethics Committee 16/SS/0154). Subsets were used to investigate relationships between inflammatory mediators in umbilical cord blood and HCA (n = 55), and between inflammation in the postnatal period and MRI markers of white matter connectivity (n = 71).
Analyte concentrations were calculated from the calibrator curves on each plate using 4PL logistic regression using the MSD Workbench software.

Analytical characterization
Intra-assay variations were calculated from 16 measurements of a pool of the same control sample on the same plate. Inter-assay variations were calculated from controls analyzed in duplicate on each plate during the sample analysis, 4 plates in total. Limit of detections were calculated as 2.5 standard deviations from duplicate measurements of the zero calibrator. The higher detection limit was defined as the highest calibrator concentration. The median intra-assay variation was 8.2% and median inter-assay variation was 11.1%. The inter-assay variation was largely influenced by CRP, RANTES and MMP-9. Detection limits and assay variations are shown in Supplementary Table S1.

Placental histopathology
Placental examination was performed by an experienced perinatal pathologist (M.J.E.) and placental reaction patterns were reported according to the site and degree of inflammation, using a structured system (Redline et al., 2003). HCA was defined as the presence of an inflammatory response in the placental membranes of any grade or stage.
Structural images were reported by a pediatric radiologist with experience in neonatal MRI (A.J.Q.) using an established system (Leuchter et al., 2014). Images with evidence of post-hemorrhagic ventricular dilatation, cystic periventricular leukomalacia or central nervous system malformation were excluded.

MRI analysis
Diffusion MRI volumes were denoised using a Marchenko-Pastur-PCA based algorithm Veraart et al., 2016). Eddy current, head movement and EPI geometric distortions were corrected using outlier replacement and slice-to-volume registration (Andersson et al., 2003(Andersson et al., , 2017Jenkinson et al., 2012) and bias field inhomogeneity correction was applied (Tustison et al., 2010). A template was constructed using data from 50 term born infants using DTI-TK and all the subjects were aligned Zhang et al., 2006). The water diffusion tensor derived maps of each subject were calculated after registration. The NODDI maps were calculated in the subject's native space with the NODDI-Bingham model using cuDIMOT (Hernandez-Fernandez et al., 2019;Tariq et al., 2016) and then the intracellular volume fraction [NDI] was propagated to the template space using the previously calculated transformations. The main skeleton of the fractional anisotropy (FA) template was created by thresholding at 0.15, and individual FA maps were projected onto this skeleton. Using this projection, the remaining maps were also projected to the white matter skeleton. A custom mask was created by editing the skeleton mask to remove CSF and grey matter contaminated areas, and by removing tracts passing through the cerebellum, the brainstem and subcortical grey matter areas. The resulting skeletonized maps were then multiplied by the custom mask. PSMD and PSNDI were calculated as the difference between the 95th and 5th percentiles on histogram analysis . Fig. 1 summarizes the pipeline for calculation of histogram based metrics.

Statistical analysis
Participant characteristics were compared using Student's t-test or Mann-Whitney U to compare distributions, and Chi-square tests were used to compare proportions. To investigate group differences in immune mediator profiles between those with and without HCA we used the Mann-Whitney U test and Bonferroni correction for multiple tests. Associations that remained significant after Bonferroni correction were analysed in logistic regression models where HCA was the dependent variable, analyte concentration was the independent variable, and GA at birth was entered as a covariate. The predictive power of those analytes that were associated with HCA after adjustment for GA at birth was determined using receiver operator characteristic (ROC) analysis.
To investigate relationships between systemic inflammation and white matter microstructure, principal component analysis (PCA) was used to identify blood analytes contributing to variance in the postnatal inflammatory profile. Analytes that contributed to PCs with eigenvalues > 1 were entered as independent variables in multivariable linear regression models with PSMD and PSNDI as dependent variables, and GA at birth and scan as covariates.
For all analyses, analytes with values less than the lower limit of detection (< LOD) were assigned the lowest detectable level prior to statistical analysis, and analytes with concentrations < LOD in ≥75% of participants were excluded. Statistical analyses were performed using SPSS version 24.0 (IBM Corp., Armonk, NY), with the exception of PCA, which was performed using R version 3.6.1 (R Core Team, 2019).

Participants
One hundred and two preterm infants (born before 32 weeks of gestation) took part, and their clinical characteristics are shown in Table 1. Fifty-five infants had umbilical cord blood obtained immediately following delivery, and seventy-one infants had blood taken on postnatal day 5 and had MRI performed at term-equivalent age.

Umbilical cord blood inflammatory profile associated with histologic chorioamnionitis
Of 55 infants with umbilical cord blood sampling and placental G. Sullivan, et al. Brain, Behavior, and Immunity 90 (2020) 311-318 histopathology, 24 (44%) were exposed to HCA during fetal life. Infants with HCA exposure had lower GA at birth than infants without HCA: mean GA 28 +0 weeks versus 29 +5 weeks (p = 0.002). There were no statistically significant group differences in birthweight, infant sex, or exposures to antenatal corticosteroids or magnesium sulphate for fetal neuroprotection.

Postnatal inflammation and white matter microstructure
We investigated whether specific components of the perinatal immune profile are associated with white matter microstructure (PSNDI and PSMD), by studying a subset of infants who had their inflammatory profile sampled on postnatal day 5 and brain MRI at term-equivalent  Sepsis: Positive blood culture with a pathogenic organism, or blood culture positive for coagulase negative staphylococcus / negative and treatment course for ≥5 days. Necrotizing enterocolitis: medical treatment for ≥7 days or surgical treatment. Bronchopulmonary dysplasia: the need for supplemental oxygen therapy or respiratory support at 36 +0 weeks gestational age. Retinopathy of prematurity: requiring treatment with laser therapy. age.
Of these 14 analytes, IL-8 was the only one to be associated with PSNDI (β = 0.221, p = 0.037), in a multivariable model that explained 48% of the variance in PSNDI at term-equivalent age (F (16,54) = 5.09, R 2 adjusted = 0.48, p < 0.001) (Fig. 3, Supplementary Table S5). None of the remaining 13 analytes were associated with PSNDI and there were no significant associations between any analyte and PSMD.

Discussion
By combining data from placenta, blood and brain MRI we have shown that IL-8 dysregulation may link systemic inflammation during perinatal life with altered white matter development in preterm infants. We found that umbilical cord blood IL-8 concentration was strongly predictive of histologic chorioamnionitis, which is an intrauterine inflammatory condition and leading cause of preterm birth, and that elevated IL-8 in the first week of postnatal life is associated with white matter dysmaturation at term-equivalent age.
IL-8 is a member of the CXC chemokine family and a mediator of the systemic inflammatory response, where its key role is neutrophil chemotaxis. IL-8 can be secreted by innate and adaptive immune cells and a variety of CNS cells including astrocytes and microglia (Choi et al., 2014;Rustenhoven et al., 2016). IL-8 binds to target cells via the Gprotein coupled receptors CXCR1 and CXCR2. These receptors are expressed throughout the CNS and are associated with key Table 2 Concentrations of analytes measured from dried blood spots taken from umbilical cord blood of preterm infants with and without exposure to histologic chorioamnionitis.  Fig. 2. ROC curve analysis of cord blood inflammatory markers for the prediction of histologic chorioamnionitis. G. Sullivan, et al. Brain, Behavior, and Immunity 90 (2020) 311-318 neurodevelopmental processes including cell migration, proliferation and differentiation, oligodendrocyte maturation, axonal growth and synaptic plasticity (Deverman and Patterson, 2009;Semple et al., 2010;Ubogu et al., 2006;Watson et al., 2020). This suggests multiple potential mechanisms through which IL-8 dysregulation during a critical window of neurodevelopment may disrupt healthy CNS development. IL-8 dysregulation has been implicated in the pathology of several types of human brain injury across the life course. It is associated with blood brain barrier dysfunction in adult traumatic brain injury (Kossmann et al., 1997;Obermeier et al., 2013), with altered cerebral metabolism and poor neurodevelopmental outcome following neonatal hypoxic ischemic injury (Bartha et al., 2004;Foster-Barber et al., 2001) and with increased mortality in children following traumatic brain injury (Woodcock and Morganti-Kossmann, 2013). Studies focused on the perinatal period report that elevated IL-8 in blood sampled from the umbilical cord or soon after birth is associated with overt white matter injury and cerebral palsy, neurodevelopmental impairment, and cognitive impairment in children born preterm, and it is one of the neonatal cytokines that is most strongly associated with subsequent diagnosis of autism among children born at term (Carlo et al., 2011;Hansen-Pupp et al., 2008;Heuer et al., 2019;Kinjo et al., 2011;Kuban et al., 2017Kuban et al., , 2014Leviton et al., 2019;Silveira and Procianoy, 2011). Furthermore, studies investigating the maternal immune activation hypothesis and offspring mental health report that elevated maternal IL-8 during pregnancy is associated with alterations in brain structure and an increased risk of schizophrenia in offspring (Brown et al., 2004;Ellman et al., 2010). This accumulating evidence suggests that inflammation associated with elevated IL-8 has a significant adverse impact on the developing brain with pervasive effects on both structure and function.
To our knowledge, this is the first study to integrate information about systemic inflammation from placenta and blood with biomarkers of white matter microstructure in a representative group of preterm infants without major parenchymal brain injury. We selected a large number of inflammation-associated proteins that are important in the innate and adaptive immune response in the newborn, and used a data driven approach based on PCA to characterize the inflammatory profile associated with HCA and white matter microstructure. Our choice of image features was principled, based on established characterizations of white matter dysmaturation in preterm infants, namely water content and dendritic/axonal complexity and dysmaturation within the white matter skeleton Kunz et al., 2014;Lynch et al., 2020).
A limitation of the study is that the cytokine response is governed not only by environmental insults but also by genetic factors and maternal inflammatory status (Dammann and O'Shea, 2008;Holst and Garnier, 2008;Sheikh et al., 2016). The sample size was not sufficient to perform sub-group analyses based on gestational age or examine potential risks or resilience conferred by the genome, and we did not have data to permit analysis of maternal cytokines. Instead, we characterized antenatal inflammation using placental tissue, which is most closely associated with fetal inflammatory responses (Gotsch et al., 2007). We chose to use a global measure of white matter microstructure because microstructural properties are substantially shared across the major white matter tracts of the newborn brain (Telford et al., 2017), and we hypothesised that systemic inflammation would likely exert a global effect; different methods would be required to investigate hypotheses about tract specific or regional susceptibilities to immune dysregulation.

Conclusions
This study provides further support for a substantial role of generalized inflammation in the etiology of preterm brain injury, and it suggests that IL-8 dysregulation may provide a link between systemic inflammation and brain dysmaturation. Further work to assess causation is warranted, such as elucidating cell-specific effects of IL-8 on developing neurons and glia in model systems. Given the potential role of immunomodulatory therapies for treating diseases of the central nervous system (Çakici et al., 2019;Wittenberg et al., 2020), and the availability of anti-IL-8 monoclonal antibody therapies, this finding has implications for the development of perinatal neuroprotection strategies based on anti-inflammatory and novel immune therapeutics.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Fig. 3. Partial regression plot of day 5 IL-8 and PSNDI, controlling for GA at birth and GA at scan. G. Sullivan, et al. Brain, Behavior, and Immunity 90 (2020) 311-318