Main

Neurotrophins (NTs)—nerve growth factor, brain-derived neurotrophic factor (BDNF), NT-3, and NT-4/5—are a family of growth factors that play an important role in the growth and development of central and peripheral nervous systems (1,2). NTs promote the growth, survival, proliferation, and migration of neurons; regulate neurotransmitter synthesis and secretion; and the development of synaptic plasticity. In addition, they also modulate immune cells (3–6). NTs are also protective against apoptotic neuronal loss, and exogenous BDNF administration has been shown to attenuate various forms of brain injury in both animals and humans (7–9).

BDNF is synthesized in several neuronal and glial cell populations and also expressed in several nonneural tissues, such as immune cells and the vascular endothelium, but the contribution of the latter to the circulating pool is limited (10–14). NTs cross the blood-brain barrier that is relatively immature in the preterm infants (15,16). Karege et al. (17) have shown that changes in serum BDNF concentrations are similar to changes in BDNF concentrations in the brain. In addition to its local effects in the central and peripheral nervous systems, BDNF plays important roles in several nonneural tissues (18–21). For example, exogenous BDNF administration has been shown to modulate glucagon secretion and glucose homeostasis (22). Changes in BDNF concentrations also correlate with disease course and response to therapy under various clinical conditions (6,23–26).

Prematurity is associated with increased morbidity and mortality and outcomes are influenced by both antenatal and postnatal factors (27–30). Preterm brain cortical volume grows nearly 4-fold between 29 and 35 weeks of gestation and is a critical period for neuronal growth and migration. Infants born at this gestation period have been shown to have decreased cortical gray and white matter at term equivalent—secondary to apoptosis and neuronal atrophy—that impacts on long-term neurodevelopmental outcomes (31–33).

There is little data regarding BDNF during this critical period of neuronal growth, especially in preterm infants. BDNF concentrations are lower in preterm infants compared with term and both are decreased compared with adults (34–36). Cord blood (CB) BDNF concentrations increase with gestational age (GA) and with use of antenatal steroids (ANS), whereas low concentrations have been associated with intraventricular hemorrhage (IVH) (36). BDNF concentrations in cerebrospinal fluid are increased in infants with perinatal depression and meningitis (37,38). In both animal models and humans, BDNF concentrations are decreased in infections (39,40). There is little data in preterm infants regarding changes in BDNF concentrations after birth, or its relationship with factors that influence neurodevelopmental outcomes. Because ANS improve neurodevelopmental outcomes (28) and increase BDNF concentrations (36), and postnatal steroids (PNS) are detrimental to the developing brain (41), we hypothesized that BDNF concentrations would be decreased by PNS use. In this study, we measured serial BDNF concentrations from birth until discharge in preterm infants <32-wk gestation. Further, we sought to correlate BDNF concentrations with other factors, such as birth weight and gestational age, and with clinical morbidities, such as IVH, bronchopulmoary dysplasia (BPD), and necrotizing enterocolitis (NEC), which are known to influence outcomes after premature birth.

MATERIALS AND METHODS

The study was approved by the University of Kentucky (UK) Institutional Review Board and written informed consent was obtained from all parents. Parents were free to withdraw their infant from the study at any time during their infant's stay in the neonatal intensive care unit (NICU).

Study design and population.

The study was conducted prospectively over a 1-y-period at the UK NICU. Fifty infants <32 wk were enrolled. Infants with known congenital anomalies or in the absence of informed consent were excluded.

Data collection and analyses.

CB was collected aseptically at delivery as reported previously (36). Serial samples were collected on day 2 (D2), between days 6 and 10 (D6), at day 30 (D30), and at day 60 (D60) of life. At the time of medically indicated blood draw, an additional 0.5 mL blood was collected, serum separated, and stored at −80°C until analyses.

BDNF assay was determined using DuoSet ELISA kits (R&D Systems, Minneapolis, MN) according to manufacturer's instructions. The minimum detection limit of this assay is 23.4 pg/mL (sensitivity) and there is no cross reactivity with other NTs, i.e. β-nerve growth factor, GDNF, NT-3, or NT-4 at 50 ng/mL. The interassay and intraassay coefficient of variation were 8.0 and 6.2%, respectively. Preliminary assays were done to determine ideal dilutions to achieve assay results within the standardized range. All samples were run in triplicate.

Clinical data.

Antenatal data including maternal age, gravida, race, smoking, chorioamnionitis (42), premature rupture of membranes (ROM) (43), duration of ROM, and antenatal steroid use for lung maturation (defined as complete if two doses of betamethasone 24 h apart or four doses of dexamethasone every 12 h in a 48-h period were given before delivery, or partial or none) (44); and neonatal characteristics of birth weight, gender, and GA (45,46) were recorded. Clinical morbidities of grades of IVH (47), BPD (oxygen dependency at 36-wk postmenstrual age) (48), NEC (Bell's stage 2 or higher) (49), retinopathy of prematurity (ROP) (50), and sepsis (defined by positive blood culture), and the use of systemic PNS were recorded.

Statistical analyses.

Descriptive analysis was used to describe changes in BDNF concentrations at different time points, and median, interquartile range, reported. Wilcoxon test was used to test the differences in BDNF concentrations between clinical outcome groups. Pearson and Spearman correlation were used to describe the relationship between BDNF concentrations at different time points and clinical outcomes. Two-way analysis of variance with contrast analyses was used to determine the changes in BDNF concentrations along time, and the difference among the time points and clinical outcomes. Multivariate analyses were carried out to determine significant clinical factors for BDNF concentration changes with time. All data were analyzed using SAS 9.0 statistical software. A p value <0.05 was considered significant. BDNF concentrations are reported as median (25–75th interquartile range).

RESULTS

Outcomes of enrollment.

Of the 50 infants enrolled, seven died before the end of first week of life, three gave consent only for CB collection, and four parents withdrew consent after initial samples were drawn. In three infants, serial samples could not be collected at appropriate times and therefore excluded from analyses.

Missing data.

CB samples were not always available as some infants were transferred after birth. In some, clinical condition (six infants on D2, five on D6), technical reasons (one on D6, one on D30), and discharge home (two on D30 and 10 on D60) precluded sample collection. We present data from patients where at least three serial samples beyond the first week of life were available (n = 33). Samples were available for CB (n = 22), D2 (n = 27), D6 (n = 27), D30 (n = 30), and D60 (n = 23).

Serial changes in BDNF.

The clinical and demographic features are presented in Table 1. BDNF concentrations were 1099 (374–2035), 46 (23–347), 1857 (678–2721), 1957 (1271–2669), and 2339 (941–3425) pg/mL in CB, on D2, D6, D30, and D60, respectively.

Table 1 Antenatal clinical features and postnatal outcomes of the study population
Full size table

After birth, there was a transient decline in BDNF concentrations followed by recovery between 6 and 10 d of life (Fig. 1). BDNF concentrations on D2 were significantly lower than CB (p = 0.03), and at D6, D30, and D60 (p < 0.001 for all comparisons). BDNF concentrations recovered by D6 (p = 0.10) and continued to rise at D30 (p = 0.06) and D60 (p = 0.01) compared with CB. There were no differences between D30 and D60 BDNF concentrations.

Figure 1
figure 1

Serial changes in BDNF concentrations in preterm infants <32-wk gestation. Box plots showing that BDNF concentrations decline at D2 compared with cord blood but continue to rise thereafter. BDNF concentrations at D2 were significantly lower than at all other times (*p = 0.03, **p < 0.001 for all comparisons). Data are shown as median (25–75th interquartile range).

Full size image

BDNF and antenatal factors.

CB BDNF concentrations correlated with duration of ROM (r = 0.43, p = 0.04). CB BDNF concentrations (but not at other time points) were significantly higher in infants who received a complete course of ANS (n = 16) compared with those (n = 6) who had received a partial course of ANS [1461 (553–2064) versus 281 (171–536) pg/mL, p = 0.04, Fig. 2]. CB was not available for the three infants who had received no ANS.

Figure 2
figure 2

Effect of ANS on BDNF concentrations. Box plots showing BDNF concentrations at various time points in infants who received ANS (white) and those who did not receive ANS (gray). Data are shown as median (25–75th interquartile range). The dotted line in the box plots represents the mean value (*p = 0.04).

Full size image

CB BDNF concentrations did not correlate with birth weight, GA, or any other antenatal or maternal factors.

BDNF and postnatal outcomes.

CB BDNF concentrations did not correlate with IVH (r = −0.19, p = 0.40) although they were lower in infants who developed IVH (n = 7) compared with those who did not develop IVH [n = 12, 694 (344–1337) versus 1473 (375–2057) pg/mL; p = 0.36]. IVH developed in 12 infants (grade I–II) with no infants with grades III or IV hemorrhage. Nine of 12 infants who developed IVH had received ANS, whereas the three infants who had not received any ANS all developed IVH.

BDNF concentrations were negatively correlated with the use of PNS (r = −0.53, p = 0.002) at D30 and at D60 (r = −0.55, p = 0.009).

In infants who received PNS (n = 7) for either hypotension or chronic lung disease, BDNF concentrations were significantly decreased at D30 and at D60 (Fig. 3). BDNF concentrations at D30 were 733 (101–1983) versus 2224 (1677–4400) pg/mL (p = 0.004) and at D60 were 1149 (288–2270) versus 2560 (1337–5166) pg/mL (p = 0.01) in infants who did and did not receive PNS, respectively.

Figure 3
figure 3

Effect of PNS on BDNF concentrations. Box plots showing that PNS use was associated with reduced BDNF concentrations at D30 (*p = 0.004) and D60 (**p = 0.014). Data are shown as median (25–75th interquartile range).

Full size image

BDNF concentrations at D60 in infants who developed ROP (n = 16) versus those who did not develop ROP (n = 7) were 1417 (553–2540) versus 3593 (2620–7433) pg/mL, respectively (p = 0.005, Fig. 4). BDNF concentrations were significantly lower in those who required surgical intervention for stage 3 or higher ROP (n = 5) compared with those with no ROP [906 (538–2489) versus 3593 (2620–7433) pg/mL, respectively, p = 0.02]. In infants with ROP, BDNF concentrations in those who required surgical intervention trended lower than those who did not require surgical intervention [906 (538–2489) versus1417 (553–2540) pg/mL, p = NS].

Figure 4
figure 4

BDNF concentrations in infants with ROP at D60. Box plots showing that BDNF concentrations are reduced in infants with ROP (p < 0.005) and in infants with surgical ROP (p = 0.02) compared with infants with no ROP. Data are shown as median (25–75th interquartile range).

Full size image

BDNF concentrations did not correlate with clinical outcomes of BPD, sepsis, and NEC.

Multivariate analyses.

To determine significant factors that affected BDNF concentrations, multivariate analyses was carried out including each of the following covariates: GA, birth weight, smoking, ANS, IVH, BPD, ROP, NEC, sepsis, and PNS. In this model, both ANS (p = 0.02) and PNS (p = 0.04) and ROP (p = 0.02) were identified as significant factors.

DISCUSSION

In this study, we have found that BDNF concentrations were correlated with factors known to be associated with deleterious developmental outcomes in preterm infants. To our knowledge, this is the first report of measured serial BDNF changes from birth until discharge in preterm infants less than 32-wk gestation.

BDNF is synthesized in several neuronal populations in central and peripheral nervous systems, vascular endothelium and immune cells (12,13,51). Endogenous fetal synthesis, maternal passage, and placenta serve as source of BDNF in the fetus, although maternal and amniotic fluid levels of BDNF fall with advancing GA (36,39,52–55). Abrupt removal of these sources could cause the sharp decline in BDNF levels after birth. The pattern of transient decline followed by rising levels is similar to that reported in term infants where BDNF concentrations fell on day 1 before rising on day 4 after birth (35), although in another study involving term and preterm infants, BDNF concentrations rose on day 1 in both groups but fell on day 4 only in preterm infants (34). In neither of these studies, BDNF concentrations were followed beyond the first week of age.

Use of ANS for pulmonary maturation are associated with improved neurodevelopmental outcomes in preterm infants (28). CB BDNF concentrations in this study were also significantly higher in infants whose mothers received a complete course of ANS (36). ANS-mediated neuroprotection may be through increased BDNF synthesis, or alternately, neuronal maturation after ANS may result in increased BDNF concentrations (36). Our data also suggests that the influence of ANS may last beyond birth, as BDNF levels were higher after the first week of age in infants who received a complete course. There was also no correlation between CB BDNF and IVH as reported previously (36). The higher antenatal steroid use and lack of any grade III or IV IVH in this study may explain these differences.

In contrast to ANS, PNS use was associated with significant attenuation in BDNF concentrations. PNS used for treatment of chronic lung disease has declined because of their detrimental effect on cortical brain growth (41). Corticosteroids inhibit cellular proliferation in the subventricular zone, down-regulate BDNF synthesis, and decrease BDNF expression in the developing brain and after ischemic injury (56–59). BDNF expression is also attenuated in conditions of neuronal atrophy or cell death and aging (60). Because BDNF crosses the blood-brain barrier (14) and peripheral BDNF levels reflect central levels (17), lower BDNF levels associated with PNS could be a reflection of either a direct suppressive effect or neuronal apoptosis. BDNF also regulates neuronal migration (61) and Reelin expression—a metalloprotein that influences the “inside-out” layering during cortical migration (62). Similarly, BDNF is involved in synaptic development and plasticity associated with learning and memory (63). Alterations in BDNF concentrations after postnatal steroid use could therefore possibly affect migration of cortical neurons during this period of brain growth.

BDNF is expressed in the visual cortex and retina during development and plays an important role in development of visual plasticity (64,65). We noted that BDNF concentrations were decreased in infants who developed ROP and trended lower in those with more severe ROP. ROP is characterized by abnormal neovascularization in which local concentrations of VEGF play a significant role in its pathogenesis (66). Conflicting data suggest that serum VEGF concentrations may be decreased or unchanged with severe ROP (67,68). BDNF is expressed in the vascular endothelium, is involved in angiogenesis, and increases VEGF expression (69,70). The presence of BDNF in the retina and its possible role in angiogenesis suggests that BDNF may play an as yet unappreciated role in the development of ROP. It remains to be determined whether low BDNF concentrations are a marker for severe ROP.

BDNF concentrations were positively correlated with duration of ROM in our study, which to the best of our knowledge has not been reported previously. Premature ROM increases risk of infections and can be considered a form of stress for the fetus (43). Alterations in expression of neuronal markers including decreased BDNF concentrations have been reported in animals exposed to prenatal stress and infections (39,71,72). However, only four infants were diagnosed with chorioamnionitis that precluded further analyses in this study. The correlation of BDNF concentrations and duration of ROM may perhaps be related to exposure to ANS the majority of infants received before delivery.

The strengths of our study include the serial measurement of BDNF concentrations in preterm infants. Our results support strong correlation between BDNF concentrations and antenatal steroid use and postnatal factors of postnatal steroid use and severe ROP, all of which are known to influence neurodevelopmental outcomes.

In summary, BDNF concentrations rise following a transient decline after birth. BDNF concentrations were strongly correlated with several antenatal factors and postnatal outcomes that significantly impact neurodevelopmental outcomes in preterm infants. A larger study and developmental follow up of these infants would be required to further answer these questions.