Proinflammatory Endothelial Phenotype in Very Preterm Infants: A Pilot Study

Very preterm infants are exposed to prenatal inflammatory processes and early postnatal hemodynamic and respiratory complications, but limited data are available about the endothelial effect of these conditions. The present pilot study investigates the perinatal endothelial phenotype in very preterm infants (VPIs) and explores its predictive value on neonatal mortality and hemodynamic and respiratory complications. Angiopoietin 1 (Ang-1), Ang-2, E-selectin, vascular adhesion molecule 1 (VCAM-1), tissue factor (TF), and endothelin 1 (ET-1) concentrations were tested in first (T1), 3rd (T2), and 7–10th (T3) day of life in 20 VPIs using Luminex technology and compared with 14 healthy full-term infants (FTIs). Compared to FTIs, VPIs had lower Ang-1 at T1 and T2; higher Ang-2 at T1, T2, and T3; higher Ang-2/Ang-1 ratio at T1, T2, and T3; lower E-selectin at T1, T2, and T3; higher VCAM-1 at T1; higher TF at T2. No differences in concentrations were found in neonatal deaths. VPIs with hemodynamic or respiratory complications had higher Ang-2 at T3. Perinatal low Ang-1 and high Ang-2 associated with high VCAM-1 and TF in VPIs suggest a proinflammatory endothelial phenotype, resulting from the synergy of a pathological prenatal inheritance and a premature extrauterine transition.


Introduction
Premature birth is always the result of a pathological process, inflammatory or associated with dysfunctional placentation [1]. Hence, premature infants are at risk of multiple early complications, and neonatal outcomes are closely related to the gestational age (GA) at delivery [2].
The endothelium constitutes one of the most extensive internal surfaces of the body and can be considered conceptually a "systemically disseminated organ" [3]. It interacts with the soluble mediators of the inflammatory response and regulates vascular permeability, coagulation balance, and angiogenesis. Moreover, endothelium modulates the vascular tone and the transport of substances between the intra-and extravascular compartments [4]. The characterization of the endothelial profile at birth could provide valuable information about critical premature neonates and their antenatal history. Indeed, the maternal processes leading to prematurity are classified into endotypes involving different molecular pathways [5,6]. Therefore, different endotypes may determine specific endothelial phenotypes. Although neonatal and pregnancy diseases have shown in-vivo and ex-vivo alterations in numerous soluble endothelial markers, including adhesion molecules, tissue factor (TF), and endothelin 1 (ET-1), currently, little is known about the endothelial profile of preterm infants [7][8][9][10][11]. Understanding it in these patients is highly complex. The immature endothelium is subjected to continuous dynamic events during the fetal and postnatal period, influenced by gestational age, oxygen blood pressure, and placental vascular growth factors, according to the process named "endothelial developmental physiology" [12][13][14].
A family of endothelial growth factors, called angiopoietins, is gripping in this context. Initially discovered as regulators of angiogenesis, angiopoietins are competitors for the transmembrane tyrosine-protein kinase receptor Tie-2 (Tie2) expressed on the luminal endothelium [15]. Growing evidence has delineated the significant role of Angiopoietin-Tie2 related endothelial activation in acute lung injury (ALI), systemic inflammatory response syndrome (SIRS), and multiorgan dysfunction in adult patients and full-term newborns [16][17][18]. Furthermore, angiopoietin 1 (Ang-1) and angiopoietin 2 (Ang-2) play an antagonistic role in regulating the inflammatory response [3]. Indeed, plasma imbalance of Ang-2/Ang-1 ratio toward the proinflammatory Ang-2 has shown a predictivity for mortality in critically ill adult, and pediatric patients admitted to intensive care units [19].
Hence, to explore the role of these pathways in preterm birth, we designed a pilot trial on a cohort of very preterm infants (VPIs, <32 weeks of gestation). The study's primary objective was to describe the profile of six soluble endothelial markers in the first 10 days of life and compare it with a control group of healthy full-term infants (FTIs). Secondary objectives were to evaluate the association between the profile of the endothelial markers and: (1) neonatal mortality; (2) early respiratory or hemodynamic complications. Our hypothesis was that VPIs had an unbalanced endothelial phenotype, whose trend is related to early outcomes.

Materials and Methods
This prospective observational pilot study was carried out from March to July 2021 at the III • level neonatal intensive care unit (NICU) of Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico, Milan, Italy. All procedures were carried out according to the principles of good clinical practice, the World Medical Association Helsinki Declaration, and the national legislative and administrative provisions in force. The study was approved by the local ethics committee (Milan Area 2, Italy) on 2 March 2021, with approval number 8997_2021. Furthermore, written informed consent to be part of the study and publication of the results was obtained from their parents.

Study Population
All consecutive VPIs were assessed at birth for eligibility. Exclusion criteria were: (1) prenatal hematological disorders, including intrauterine transfusions; (2) perinatal asphyxia, defined as an Apgar score at 10 ≤ 5, or any pH ≤ 7.0 or base deficit ≥ 12 mmol/L in the first hour of life [20].
As control group, we evaluated all consecutive healthy FTIs (≥37 weeks of gestation) rooming-in hospitalized with indications to perform blood tests on the first day of life just for obstetric risk factors. FTIs with clinical signs suggestive of infection, respiratory distress, or hemodynamic instability at the sample test were excluded from the study.
In addition, as per department protocol, all VPIs received cardiac and cerebral ultrasound surveillance on days 1, 3, and 7. For both groups, we collected demographic data related to maternal medical history during pregnancy (multiple pregnancies, drugs, fetal growth, preeclampsia, chorioamnionitis), perinatal (gender, delivery mode, Apgar, cardiopulmonary resuscitation), and postnatal period (prematurity-related comorbidities, clinical course and outcome of blood tests in FTIs).
The whole blood (0.5 mL) was collected in an ethylenediaminetetraacetic acid (EDTA)containing test tube for each evaluation. The samples were centrifuged at 3000 rpm at 4 • C for 15 min. The supernatant was collected and stored at −20 • C. According to the manufacturer's protocols, the analysis was performed with Luminex technology (R&D System, Inc., Minneapolis, MN, USA).
The sampling schedule was 1st (T1), 3rd (T2), and 7-10th (T3) day of life in the VPIs cohort and 1st day of life in the control group.

Prematurity-Related Outcomes
Neonatal mortality was defined as death within the 28th day of life. In addition, early respiratory and hemodynamic complications were evaluated as the occurrence in the first 72 h of any of the following: (1) Respiratory failure (need for mechanical ventilation); (2) Arterial hypotension (need for vasopressor drugs to maintain a mean arterial pressure ≥ 10th percentile for the patient's GA) [21]; (3) ultrasound evaluation of hemodynamically significant patent ductus arteriosus (PDA) [22]; (4) ultrasound evaluation of severe (grades 3-4) intraventricular hemorrhage (IVH) [23]. Finally, a post hoc analysis of subgroups divided for maternal endotypes at T1 was performed to evaluate the prenatal contribution to the endothelial phenotype. The endotypes were classified as follows: inflammation (chorioamnionitis or placental abruption) vs. dysfunctional placentation (preeclampsia or fetal growth restriction) [6].

Statistical Analysis
According to the literature, we planned a sample size of 20 VPIs and 14 FTIs for this pilot study, allowing for the dropout of technical issues and adverse clinical outcomes [24]. Demographic characteristics of the two populations were presented as mean (SD), median [IQR], or n (%). Patients who have not completed the 3 scheduled samples are still included in the descriptive population assessments and comparing VPIs and FTIs. Mann-Whitney U test was used to compare FTIs and VPIs data, dead and survived infants' values, and maternal endotype-related values. Friedman test with false discovery rate correction in post hoc tests was used to study significant variations between the 3 samples of each soluble marker; in particular, only VPIs with 3 available measurements were considered [25]. Linear regression models were used to study the association between GA or birth weight (independent variables) and soluble markers at T1 (dependent variables). Results are expressed as an estimated coefficient, 95% CI, and p-value. Statistical significance was defined for p-values < 0.05.

Endothelial Profile
The assessments of soluble endothelial markers were 67: 53 from the VPIs group and 14 from the control group. In this case 53 VPIs group observations are subdivided as follows: 20 (100%) at T1, 17 (85%) at T2, and 16 (80%) at T3. Defections are due to 4 deaths in the first week of life. ET-1 values were undetectable in all patients of both groups using the measurement technique. The complete results of the other soluble endothelial markers are summarized in Table 2. Higher Ang-2 at T1, T2, and T3 (p < 0.001); 3.
The relationship between soluble endothelial markers of VPIs and FTIs is described in Figure 1.

Relationship between T1-T2-T3
Friedman test showed significant changes in the first 10 days of life in the soluble markers analyzed. The relationship between soluble endothelial markers at T1-T2-T3 is described in Figure 2.

Neonatal Mortality
Mortality analysis was performed by comparing only the observations at T1 due to deaths before T2 collection of 3 out of 5 infants (60%). There were no significant differences in soluble makers between the 15 surviving and 5 dead VPIs ( Table 3). The soluble markers showed the following mean values in dead compared with survivors VPIs: Ang-1 48% lower, Ang-2 24% lower, Ang-2/Ang-1 58% higher, E-selectin 23% higher, VCAM-1 16% higher, TF 139% higher. The dead patients showed significantly lower gestational age

Discussion
To our knowledge, this pilot study first explores in-vivo a perinatal endothelial profile of VPIs compared to a control group of FTIs. Previous studies have evaluated soluble angiogenic factors and markers of vascular contractility in preterm infants with bronchopulmonary dysplasia and hypoxic encephalopathy, respectively, but none of these assessed angiopoietins balance [11,29,30].
We observed significant differences in angiopoietins concentrations between VPIs and healthy FTIs, configuring a higher Ang-2/Ang-1 ratio from birth to the 10th day of life in the VPIs group. Ang-1 promotes endothelial quiescence by enhancing cell survival and downregulating gene expression of proinflammatory and pro-coagulating pathways [31]. Conversely, rapid endothelial exocytosis of Ang-2 during hypoxic or inflammatory endothelial injury increases the surface expression of cell adhesion molecules. This destabilizes intercellular junctions with a consequent increase in microvascular permeability, which is a key mechanism in the pathophysiology of organ damage [32]. According to these biological processes, the perinatal increase in Ang-2, associated with a reduction in the antagonistic effect of Ang-1, displays inflammatory endothelial activation in premature infants.
Angiopoietins also play a fundamental role in regulating fetal angiogenesis, and our finding of a negative association between GA and Ang-2 suggests a higher physiological expression in preterms. However, knockout mouse models for Ang-1 or Ang-2 demonstrated a fine prenatal regulation in the concentrations of Tie-2 receptor competitors [33,34]. Furthermore, the fetal absence of Ang-1, similarly to the overexpression of Ang-2, inevitably leads to abortion, demonstrating that Ang-1 plays the leading role in a correct and orderly angiogenic process while Ang-2 acts as its counter-regulatory [15]. Therefore, ontogenesis alone should not justify the imbalance of VPIs towards Ang-2. Considering the Ang-2 involvement in adult conditions common to preterms, such as SIRS and ALI, we could assume that the VPIs also manifest an inflamed endothelium [17,35].
Moreover, the inflammatory hypothesis is supported by the concurrent higher levels of VCAM-1 and TF in VPIs. Ex-vivo studies have shown that human umbilical vein endothelial cells (HUVECs) of ELBW infants can generate adequate up-regulation of adhesion molecules, including VCAM-1, in response to proinflammatory cytokines [36]. In-vivo studies confirmed this finding by increasing VCAM-1 on cord blood samples of preterms born from maternal chorioamnionitis [37]. Furthermore, the interleukin-1β stimulation of HUVECs increases the subendothelial exposition of TF through the induction of vascular permeability, which Ang-2 also stimulates [38]. In this regard, the association between GA and TF could be secondary to increased permeability induced by Ang-2 in the most immature patients.
Conversely, E-selectin values showed an opposite trend compared to VCAM-1 and TF. At birth, it was lower in VPIs but tended to progressively increase at T2 and T3 without reaching FTIs' levels. This does not necessarily contradict the inflammatory hypothesis but encourages the theory of an immature expression of this protein in preterm infants [37]. Indeed, the capability of up-regulating E-selectin at very low GA is debated, and some experimental data exclude its expression before the 32nd week of GA [39]. Nevertheless, these results demonstrate that soluble E-selectin is already present at extremely low GA with a time-dependent up-regulation trend.
Two moments can induce an inflammatory endothelial phenotype in VPIs: the fetal life, depending on the maternal endotype, and the postnatal life, depending on the hemodynamic and respiratory adaptation. Understanding and quantifying the contribution of the two moments are challenging, but the exposure kinetics of the molecules involved can provide some inferences.
Ang-1 and VCAM-1 modulation depend on nuclear transcription factors and take up to 3 h to manifest [40][41][42]. This premise suggests that T1 sampling, performed shortly after birth, was only partially affected by postnatal factors and that intrauterine inheritance can play a key role in this finding. However, the origin of this inheritance is still to be explored. The relationship between inflammatory vs. dysfunctional placentation endotypes at T1 did not give detailed answers. Although there are no significant differences, the dysfunctional placentation endotype showed a consistently less proinflammatory phenotype for all parameters analyzed. Furthermore, as an additional prenatal factor, the possible effect of steroid prophylaxis (80% of our population) on the endothelial phenotype should be evaluated for future studies.
Considering the postnatal contribution to the endothelial phenotype, a strength of the study was the prospective enrollment of critically ill patients, 75% of whom were ELBW infants, with high complication rates. The Ang-2 increase at T2, with no decrease at T3, is significant in this setting. Furthermore, Ang-2 is stored in preformed stocks in Weibel-Palade bodies, allowing a concentration peak as early as 30 min after the inflammatory stimulus, associated with a marked reduction of its blood concentrations as early as 60 min [43]. Therefore, a postnatal stimulus crosses the first days of life of VPIs, maintaining a prevalent Ang-2 phenotype. This suggests that premature birth is a stressful event for the endothelium or probably that the endothelium reflects hemodynamic and respiratory stress. The late Ang-2 prevalence in VPIs suffering from early complications, associated with a proinflammatory trend already shown at T1 by this group, supported this relationship.
Conversely, the increase of Ang-1, associated with VCAM-1 and TF reduction, suggests an anti-inflammatory counter-regulation in the late transition phase. On the 10th day of life, hemodynamic and respiratory adaptation is a subacute process for infants who are not affected by severe complications, and this could correlate with a re-balancing of the Ang-2/Ang-1 ratio. Accordingly, a protective role of Ang-1 has been proposed, which would be consistent with the trend of Ang-1 shown by deceased newborns, even if not statistically significant. [17].
The findings of this study introduce new evidence on the perinatal endothelium of preterm infants, but some limitations require consideration. This is an exploratory pilot study, and the small sample size, especially of subgroups, only provides a trend that is limited in showing a predictive value of angiopoietins on mortality and early acute complications. In addition, by ethical practice, the control group is limited to a single observation performed on the first day of life and not to serial samplings as for the group of VPIs that would better show the postnatal trend in healthy term infants. Finally, the analytical method showed insufficient sensitivity to detect soluble ET-1 in both the VPIs and the control groups. The evaluation of vascular contractility-related markers remains an area yet to be explored in prematurity, as the few data available on the nitric oxide (NO), endothelial nitric oxide synthase (eNOS), and ET-1 levels have shown controversial results [10,[44][45][46][47]. However, clinical evidence of impaired modulation in preterm infants' systemic and pulmonary vascular contractility suggests further efforts.
In conclusion, exploring the endothelial profile in a cohort of premature infants revealed a perinatal proinflammatory attitude due to a reduced Ang-1 and an overexpressed Ang-2. The exact mechanisms responsible for this phenotype and a possible prognostic value of angiopoietins on mortality and early complications await further studies.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Acknowledgments:
A special thanks to all the nurses and medical doctors of the NICU of the Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico and Lucrezia Magnini, Andrea Carlin, and Patrizia Leonardi, who helped us in the management of the experimental analyzes.

Conflicts of Interest:
The authors declare no conflict of interest.