At the time of term birth, the immune system has not fully matured. The inexperienced adaptive immune system must still develop specificity and memory, which is completed only in the early childhood years (Hannet et al., 1992). As such, normal term neonates rely heavily on their innate immune response but this too is immature (Marodi, 2006). Immaturity of the immune system is more pronounced in infants born preterm (Figure 1).

Intrauterine inflammation is the principal cause of preterm birth (Hagberg et al., 2002; Romero et al., 2007; Fahey, 2008). Intrauterine inflammation can be caused by bacterial infection, which can occur by ascent of bacteria from the birth canal, crossing of the placenta or membranes by microbes, or inadvertent pathogen transfer into the amniotic cavity during amniocentesis (Hagberg et al., 2002; Goldenberg and Culhane, 2003). Intrauterine inflammation is inversely correlated with gestational age; it is present for 83% of infants weighing less than 1000 g at birth and for 10% in infants greater than 2500 g (Goldenberg and Culhane, 2003). Intrauterine inflammation affects as many as 30% of all neonates born at ≤34 weeks gestation (Lahra and Jeffery, 2004).

Intrauterine inflammation can increase the risk of infant mortality (Hagberg et al., 2002; Fahey, 2008). Clinical and experimental studies demonstrate that bacteria or pro-inflammatory mediators in amniotic fluid can be inspired or swallowed by the fetus to elicit a fetal inflammatory response (FIRS) (Gotsch et al., 2007), which is also observed in a sheep model of intrauterine inflammation (Moss et al., 2002c). In sheep, intrauterine inflammation increases the risk of perinatal neurological injury (Duncan et al., 2006; Nitsos et al., 2006) and impaired gut development (Wolfs et al., 2009). The fetal sheep lung has increased surfactant production and improved compliance after exposure to inflammation (Moss et al., 2002a), consistent with reduced rates of respiratory distress syndrome (RDS) in preterm human neonates exposed to inflammation in utero (Lahra and Jeffery, 2004). However, at least in the sheep, lung structure is simplified (Moss et al., 2002b). Simplified lung structure arising from prenatal exposure to inflammation could contribute to a potential increase in the risk of bronchopulmonary dysplasia (BPD) in human neonates (Been et al., 2009).

FIRS is characterized by an increase in fetal plasma IL-6, C-reactive protein, IL-1, IL-8, and GM-CSF (Berry et al., 1995; Goldenberg et al., 2000; Gotsch et al., 2007). There is evidence of monocyte and neutrophil activation in addition to increased numbers of these cells in the fetal circulation after exposure to inflammation in humans and sheep (Kallapur et al., 2007; Kramer et al., 2007; Romero et al., 2007). Lymphocytes are activated during infections in utero in humans, indicating the fetal adaptive immune response is at least partly responsive (Duggan et al., 2001). Fetuses and neonates exposed to intrauterine inflammation have increased Th1 cells corresponding with an increase in IFN-γ, indicating a potential shift from Th2 to Th1 of the fetus. The shift to Th1 cytokines may lead to membrane rupture because normal term labor is partially an inflammatory event, with an increase in the production of Th1 cytokines TNF-α, IFN-γ, IL-1β, and prostaglandins in the fetal membranes and amniotic fluid (Sykes et al., 2012). Intrauterine inflammation also increases production of these cytokines (Romero et al., 2011) and prostaglandins (Westover et al., 2012).

Animal experimentation has been valuable for understanding the immune consequences of intrauterine inflammation. Immune cells (including monocytes, neutrophils, and lymphocytes) in fetal sheep lung tissue significantly increase in response to intra-amniotic lipopolysaccharide (LPS) infusion (Kallapur et al., 2007; Kramer et al., 2007). Fetal thymic cell populations are altered after LPS exposure, resulting in a decrease in CD8 and MHC II expression on thymocytes (Melville et al., 2012); however, LPS up-regulates MHC II expression on circulating fetal monocytes (Kramer et al., 2005). The effect of LPS on MHC II appears to be tissue-dependent. In the thymus MHC II is not functioning to activate T cells, but is involved in selection and development of T cells. Thus, the reduction in MHC II may lead to altered CD4 production. Peripherally, activation of leukocytes resulting in MHC II up-regulation would aid in eliminating the threat. The increased number of lymphocytes and expression of MHC II may indicate that the fetus is capable of reacting to infection with an adaptive immune response, along with these innate responses, consistent with observations in humans (Duggan et al., 2001). Functional “maturation” of the immune system may be a consequence of intrauterine inflammation because preterm monocyte hydrogen peroxide and cytokine production increases after infusion of LPS into the amniotic cavity in sheep (Kramer et al., 2005, 2007; Kallapur et al., 2007). Responsiveness of the preterm immune system to intrauterine inflammation is demonstrated further by the observation that repeated pro-inflammatory exposures induce tolerance in preterm sheep. Repeated doses of LPS into the amniotic cavity of sheep, at 2 and 7 days before preterm delivery, cause a reduction in IL-6 secretion in fetal sheep when compared to a single dose of LPS (Kallapur et al., 2007). This tolerance effect clearly demonstrates modulation of the immune system in response to the initial stimulus.

Deficiencies in preterm immune function have implications for the eradication of postnatal infections. Particularly important are nosocomial infections, which occur more frequently in preterm infants due to their extended hospital stays. Mortality associated with early-onset sepsis is increased in preterm infants, and rates correlate inversely with gestational age. Of the infants born after exposure to chorioamnionitis, 4.8% develop sepsis (Soraisham et al., 2009) and 16% of infants with sepsis die (Adams-Chapman, 2012). Intrauterine inflammation increases the risk of early-onset sepsis, likely because at least some of these postnatal infections originated in utero. However, intrauterine inflammation decreases the risk of late-onset sepsis (Strunk et al., 2012), potentially because these infants experience immune “maturation” by the earlier (intrauterine) exposure to infection, or because they require less respiratory support and accompanying invasive care.

Antenatal glucocorticoids have been utilized clinically for prevention of neonatal respiratory disease in preterm infants for ~30 years. In 1969, Liggins used the synthetic glucocorticoid, dexamethasone, to induce preterm labor in pregnant ewes and found that the lambs born after dexamethasone were able to breathe spontaneously, whereas preterm lambs that had not received dexamethasone could not (Liggins, 1969). Since this time, numerous randomized controlled trials have demonstrated beneficial effects on the preterm respiratory, central nervous, and gastrointestinal systems of antenatal glucocorticoids (Roberts and Dalziel, 2006).

Glucocorticoids are used in adults to modify immune responses in allergy and autoimmune diseases but the equivalent effects on the fetal immune system are not well-understood (Tuckermann et al., 2005).

The hypothalamic-pituitary-adrenal (HPA) axis regulates many physiological processes, including the immune system. An increase in pro-inflammatory cytokines causes activation of the HPA axis, which increases cortisol. Cortisol suppresses the immune system, by acting on glucocorticoid receptors of leukocytes, which causes translocation to the nucleus (Beato et al., 1995) and interference with transcription of pro-inflammatory factors (NF-κB and STAT) (Tuckermann et al., 2005). Interference with these transcription factors reduces cytokine production and prevents the immune system overwhelming the body with inflammation (Barnes, 2005). Glucocorticoids can also induce apoptosis of DCs and lymphocytes (Barnes, 2005).

The effect of an increase in endogenous glucocorticoids (via stress) or maternal administration of exogenous glucocorticoids on the developing immune system has not been extensively studied in humans or other animals but the limited data show lasting effects. Glucocorticoid exposure results in long-lasting alterations to physiological and cellular responses to infection and inflammation in the offspring. In pigs, maternal cortisol treatment during pregnancy caused an increased febrile response of female offspring to LPS challenge (de Groot et al., 2007). Administration of adrenocorticotropic hormone to pregnant Rhesus monkeys over a 2-week period alters immune responses of the juvenile offspring, reducing their febrile and cytokine responses to IL-1 (Reyes and Coe, 1997). Induction of maternal stress (causing an increase in maternal corticosterone) in rats resulted in a reduction in proliferation of certain subgroups of lymphocytes after mitogenic stimulation, in offspring at 8–9 weeks postnatal age (Kay et al., 1998).

Human clinical studies have observed a reduction in lymphocyte number in preterm neonates after antenatal glucocorticoids (Chabra et al., 1998; Kavelaars et al., 1999) but there appears to be an overall increase in total leukocyte count, specifically an increase in neutrophils (Fuenfer et al., 1987; Barak et al., 1992). Meta-analysis shows that antenatal glucocorticoids may reduce the risk of neonatal sepsis (Dembinski et al., 2003; Roberts and Dalziel, 2006). However, an association between increased early-onset neonatal sepsis and multiple-course maternal betamethasone treatment has been observed in one study (Vermillion et al., 2000). A recent study showed that the ability of neonatal cord blood leukocytes to produce cytokines is not altered with antenatal glucocorticoid treatment (Kumar et al., 2011).

Intravenous administration of dexamethasone to fetal sheep, in a dose approximating that used antenatally in women at risk of preterm birth, results in an increase in total leukocytes between three and 12 h after administration due to an increase in neutrophils, but a reduction in lymphocytes (Edelstone et al., 1978). In contrast, maternal intravenous dexamethasone administration does not alter fetal circulating leukocyte counts but increases maternal neutrophils and decreases maternal lymphocytes (Edelstone et al., 1978). Similarly, pregnant women administered antenatal glucocorticoids have increased circulating neutrophils coupled with a decrease in lymphocytes (Wallace et al., 1998).

Betamethasone administration to pregnant rats on gestational days 19 and 20 (term 22 days) caused a reduction in lymphocyte proliferation and IL-2 production in late-gestation fetuses and for up to six days after birth (Murthy and Moya, 1994). Natural killer (NK) cells also had reduced cytotoxicity, particularly those from male offspring (Kay et al., 1998). Functional changes are also observed in monocytes of lambs born after maternal betamethasone injection. Phagocytosis of apoptotic neutrophils, but not bacteria, by preterm sheep monocytes was initially decreased after maternal betamethasone but was increased 3-fold seven days after (Kramer et al., 2004). These differences in phagocytic capacity may be explained by the observation that TLR-4 expression is not altered by glucocorticoids (Cepika et al., 2010), thereby not influencing detection of gram-negative bacteria. Overall, there appears to be an initial characteristic anti-inflammatory effect of glucocorticoids in the fetus, followed by “maturation” of immune function. Similar results are seen in adult monocytes exposed to glucocorticoids. Human monocytes are altered to an anti-inflammatory phenotype aiding in the resolution of infections (Tsianakas et al., 2012). In murine monocytes, glucocorticoids increase the expression of receptors responsible for the phagocytosis of apoptotic neutrophils (Nilsson et al., 2012).

From the available studies it appears likely that antenatal glucocorticoids have a modulating effect on preterm immune function. What is unknown is whether these effects are transient or persist into childhood and beyond.

Preterm infants are born before completed development and maturation of the lungs. Therefore, they have a thick blood-gas barrier, undifferentiated or immature airway epithelium and reduced ability to produce surfactant (Pringle, 1986). They also may have difficulty clearing lung liquid, resulting in lungs with a reduced ability for gas-exchange and poor compliance, making them prone to collapse (Jobe et al., 2008). Reduced surfactant production, impaired gas exchange and inefficient independent respiration can result in the need for mechanical ventilation (Brown and DiBlasi, 2011).

Mechanical ventilation of neonatal lungs can cause ventilation-induced lung injury (VILI). In preterm infants, VILI is believed to be a major contributor to the development of the chronic respiratory disease BPD. The shear stress, inspiratory volume, air pressure, and oxygen concentration of ventilation are believed to cause epithelial cell damage, which contributes to protein leak into the airways, inhibiting the function of surfactant and increasing inflammatory cell infiltration (Attar and Donn, 2002; Hillman et al., 2007; Jobe et al., 2008). The most common inflammatory cell infiltrate is neutrophils, which migrate into the airways (Jobe et al., 2008; Hillman et al., 2009). Inflammation caused by stretch injury in preterm sheep results in elevated pulmonary mRNA expression of serum amyloid A3, IL-1β, and IL-6 with 1 h of ventilation (Hillman et al., 2011). Interleukin-6 and IL-8 increase as early as 20 min after ventilation (Wallace et al., 2009).

Ventilation of the preterm lungs also results in a systemic inflammatory response. Circulating phagocytes undergo activation in mechanically ventilated infants, resulting in increasing expression of the cell adhesion molecule CD11b on both neutrophils and monocytes (Turunen et al., 2006). Activation of CD4 and CD8 T cells occurs in preterm infants with RDS requiring mechanical ventilation, and there is increasing expression of CD54 and decreasing CD62L in the first three postnatal days (Turunen et al., 2009). Lymphocyte activation is further increased in preterm infants that develop BPD (Turunen et al., 2009). However, the absolute lymphocyte number is decreased in preterm neonates with BPD, contributed to by a decrease in circulating CD4 cells (Ballabh et al., 2003).

The changes in the activation of circulating leukocytes are coupled with increased cytokine production. Term infants that require mechanical ventilation have an increased serum plasma level of IL-8, IL-1β, and TNF-α, no change in IL-6, and a decrease in IL-10 (Bohrer et al., 2010). Extended duration on a ventilator increases the magnitude of pro-inflammatory cytokine proteins detected in serum from ventilated preterm neonates (Bose et al., 2013). Preterm infants that go on to develop BPD following mechanical ventilation also have increase serum levels of TNF-α and IL-6, but decreased IL-10 (Koksal et al., 2012), indicating that sustained systemic inflammation may be a risk factor for developing chronic lung diseases.

Rates of birth by Caesarean section have risen in developed countries in recent years, such that this mode of delivery now accounts for almost one-third of births (Li et al., 2011; Martin et al., 2011). The perinatal risks of Caesarean delivery for the mother and newborn are well-appreciated but there is emerging evidence of an impact of Caesarean section on long-term immune function.

A relationship between birth by Caesarean section and increased risk of asthma was first demonstrated just over a decade ago (Xu et al., 2000, 2001) and has since been confirmed by meta-analysis of over 20 individual studies (Thavagnanam et al., 2008). Some studies have shown an increased risk for atopy in children born by Caesarean section (Maitra et al., 2004; Renz-Polster et al., 2005; Salam et al., 2006; Pistiner et al., 2008; Kolokotroni et al., 2012), leaving them at an increased risk for hypersensitivity (Pistiner et al., 2008; Kolokotroni et al., 2012) but the relationship is not straightforward. Kolokotroni et al. found an increased risk of atopy in children born by caesarean section only for families with a history of allergic disease (Kolokotroni et al., 2012).

Relationships between Caesarean delivery and increased risk of asthma and allergic disease have been attributed to differences in microbial colonization of infants at birth (Gronlund et al., 1999) and the influence of gastrointestinal flora on development of the immune system (Siggers et al., 2011). Infants delivered by Caesarean section have lower diversity of gut microflora at 3 days of age than those delivered vaginally (Biasucci et al., 2010). Recent consideration of the human body not as an individual but as a symbiotic mix of human and microbial cells, with optimal function determined by the microbiome of the host, suggests that Caesarean section may contribute to a myriad of postnatal diseases, simply by influencing gastrointestinal colonization at birth.

The systemic immune function of neonates may also be impacted by Caesarean section because more than half of Australian women who deliver in this way do so without labor (Li et al., 2011). Newborn monocyte expression of TLR-2 and -4, critical mediators of innate immunity, is reduced in the absence of labor (Shen et al., 2009), potentially reducing the responsiveness of neonates to bacteria and viruses.

Advances in obstetric and neonatal medicine have enabled profound reductions in perinatal mortality in recent years but this benefit has not come without cost. With increased rates of survival of preterm infants has come growing numbers of babies with illness and long-term disability, even for those born close to term (Cheong and Doyle, 2012). The consequences for immune development and function of preterm birth are largely unknown. We need to better understand the impact of preterm birth and associated factors, including obstetric and neonatal management, on immune function in order to improve health outcomes for the increasing number of individuals born preterm.