Neurofilaments are one of the components of neuronal axons in the brain and are detected in extracellular fluid after neuroaxonal damage [1]. Recently, neurofilaments present in Cerebrospinal Fluid (CSF) were reported to be a specific biomarker for axonal injury in infants and children [1-4]. The phosphorylated form of neurofilament H(pNf-H) may be a clinically useful neuronal biomarker because it can be measured in blood specimens [1,5-7]. Here, we encountered an Extremely Low Birth Weight Infant (ELBWI) with Posthemorrhagic Hydrocephalus (PHHC) who developed Brain Stem Release Phenomena (BRP) during continuous intravenous infusion of Midazolam (MDZ). We report the usefulness of serum pNf-H as a biomarker for neuronal damage in this ELBWI.

The patient was the first girl of a healthy non-consanguineous 33-year-old mother and 27-year-old father. She was born after 23 weeks of gestation by emergency caesarian section because of hemolysis, elevated liver enzymes, and a low platelet count. Her Apgar score was 2 at 1 min and 3 at 5 min. Birth weight was 275 g (<10^th percentile), length was 21.5 cm (<10^th percentile), and head circumference was 17.2 cm (<10^th percentile). The patient was admitted to our neonatal intensive care unit where she received mechanical ventilation, was administered dopamine, indomethacin, antibiotics, antifungal agents, and fentanyl underwent frequent transfusions of blood products. On day 6, she exhibited an invasive fungal infection complicated with disseminated intravascular coagulation and reopening of the ductus arteriosus. On day 8, she developed grade IV Intraventricular Hemorrhage (IVH) that was detected by Ultrasonography (US) (Figure 1a), for which she was monitored by continuous amplitude-integrated Electroencephalogram (aEEG). Her clinical course is shown in Figure 2. We gave her intravenous Phenobarbital (PB) for sedation and to prevent Neonatal Seizure (NS), but despite a therapeutic PB blood level of 21.5 μg/ml, she experienced her first NS that persisted for 9 hours on day 9, which consisted of tachycardia, desaturation, and tachypnea with no apparent convulsions. Readings for aEEG showed a typical saw tooth pattern, and rhythmic slow wave bursts were recorded by raw EEG (Figure 1c). The NS disappeared after intravenous administration of MDZ at a dose of 0.1 mg/kg and then continuous infusion of 0.15 mg/kg/h. On day 25, a second NS with clinical symptoms and EEG abnormalities similar to the initial attack occurred for 6 hours. Although PB was ineffective, MDZ led to the discontinuation of the NS. The blood levels of PB and MDZ were 11.3 μg/ml and 10.0 ng/ml, respectively. On day 28, she began to exhibit abnormal repetitive movements that mimicked swimming in the upper extremities and pedaling in the lower extremities, which lasted for 4 hours. Since EEG measurements showed no seizure activity (Figure 1d) and the movements could be suppressed by physical restraint, we diagnosed them as BRP. These abnormal movements were noted even though MDZ administration had been stopped for more than 2 hours. The patient’s PHHC became progressively worse (Figure 1b), and so she recommenced MDZ treatment and ultimately underwent surgery for PHHC on day 35, after which neither NS nor BRP were observed again.

Figure 1a: Ultrasonography (US) performed on day 8 shows grade IV intraventricular hemorrhage.

Figure 1b: Ultrasonography on day 28 shows posthemorrhagic hydrocephalus.

Figure 1c: Amplitude-integrated Electroencephalogram (aEEG) (upper) and raw EEG (lower) readings obtained on day 9 depict seizure activities

Figure 1d: aEEG (upper) and raw EEG (lower) readings taken on day 28 show no seizure activities.

Figure 2: Clinical course and serum pNf-H levels of the present patient. The shaded area indicates normal range (<0.94 ng/ml). Note that pNf-H levels on days 3, 8, and 29 were markedly higher than those of controls. The excessive water input was observed on day 8-10. The concentrations of methemoglobin were transiently raised on day 6-8 and gradually increased according to the progression of hydrocephalus. BRP, brainstem release phenomenon; IVH, intraventricular hemorrhage; MDZ, midazolam; PB, phenobarbital; pNfH, phosphorylated neurofilament H.

To examine the usefulness as a biomarker for neuronal injury in preterm infants, we serially measured the serum levels of pNf-H in this patient and in age-matched control infants. As a control group, 19 serum samples were obtained from very low birth weight infants; 10.9 ± 7.6 days old (range, 0-24 days), 27.8 ± 4.4 weeks of gestation (range, 23-32 weeks) and 848.8 ± 30.9 g of birth weight (range, 552- 1064 g). All of them were delivered at Shinshu University Hospital between April, 2011 and March, 2012 who were considered to be neurologically normal based on physical findings, magnetic resonance imaging, auditory brainstem response, and EEG. We measured pNf-H with a Phosphorylated Neurofilament H Sandwich ELISA Kit (Merck Millipore®, Billerica, Massachusetts) according to the manufacturer’s instructions. Briefly, chicken polyclonal antibodies against pNf-H were pre-coated onto a 96-well plate and used to capture pNf-H from a 50- μl serum sample. Retained pNf-H was detected using pNf-H-specific rabbit polyclonal antibodies and goat anti-rabbit IgG polyclonal antibody-conjugated alkaline phosphatase. After addition of the substrate solution, the amount of pNf-H was determined using an ELISA plate reader at an absorbance of 405 nm. The assay detection limit was 0.0293 ng/ml. Samples were analyzed in duplicate. Serum samples were separated immediately and kept at -30°C until analysis. Written informed consent was obtained from the parents of all subjects. This study was approved by the Ethics Committee of Shinshu University.

Serum concentrations of pNf-H are presented in Figure 2. The mean value obtained from the 19 healthy control samples of very low birth weight infants in the neonatal period was 0.43 ± 0.51 ng/ ml (mean ± SD). Serum pNf-H levels in our patient were higher than the control mean ± SD values at every time point. The first peak of pNf-H observed on day 3 was extremely high and then decreased. After IVH developed on day 8, serum pNf-H gradually increased again, apparently along with the progression of PHHC, until day 29. We also analyzed the concentrations of methemoglobin, which were transiently raised on day 6-8 and gradually increased according to the progression of hydrocephalus.

Recent reports have demonstrated the clinical usefulness of serum pNf-H in term infants with hypoxic ischemic encephalopathy [1,2], but none have addressed pNf-H in premature infants or progressive brain damages in infants. In this study, the serum pNf-H of our patient was persistently elevated and peaked 3 and 29 days after birth. The first peak was considered to be due to asphyxia at birth, and the ensuing gradual elevation appeared to coincide with the progression of PHHC. These results showed that pNf-H levels sensitively reflected neuronal damage in this patient. The rapid reduction of pNf-H level on day 9 was considered to be caused by excessive water input on day 8-10. Brainderived creatinine kinase and neuron-specific enolase are conventionally used to assess brain damage, but these molecules are also present in the placenta and blood cells. Though matrix metalloproteinase 9 and 2 are also known to be useful biomarker for neonatal intraventricular hemorrhage, these are increased in not only neuronal damage but also neonatal bronchopulmonary dysplasia [8]. In contrast, neurofilaments are specific components of the nervous system and thus represent markers for neuronal damage. While neurofilament-L (Nf-L) in the CSF of term and preterm neonates with perinatal asphyxia or PHHC has been reported as a neuronal biomarker [2], pNf-H is more resistant to calpain and other proteases [9]. In addition, pNf-H can be assayed in very small amounts (50 μL/assay) of blood samples, which is clinically important for ELBWI. Thus, monitoring serum pNf-H may be useful in estimating CNS damage in preterm infants. Furthermore, we analyzed the concentrations of methemoglobin, which were transiently raised on just before intraventricular hemorrhage and gradually increased according to the progression of hydrocephalus. Methemoglobin, which was suggested to be crucial initiator in the development of brain damage [10], also might to be the predictor of neuronal complications.

BRP in neonates shows paroxysmal automatic movements without epileptic discharges in EEG [11,12]. The major causes of BRP are cortical dysfunctions induced by IVH, hydrocephalus, and hypoxic ischemic encephalopathy. Cortical damage leads to the release of brainstem preprogram activity, such as oral-buccal movements, ocular signs, and limb and axial movements mimicking swimming or rotatory motions in the upper extremities and stepping or pedaling motions in the lower extremities [11]. BRP is also considered to be induced by the administration of AEDs, especially benzodiazepines including MDZ, which suppress cortical activity [12]. In this patient, BRP was observed on day 28 after her PHHC had gradually worsened but no MDZ was detected in the blood at that time. Thus, it can be said that her severe progressive PHHC, and not the administration of MDZ, primarily contributed to the development of BRP.

In conclusion, this is the first report to show that monitoring of serum phosphorylated form of neurofilament H levels is useful in estimating central nervous system damage and to identify the cause of BRP in ELBWI with PHHC.

This work was supported by Japanese Epilepsy Research Foundation and the Preventive Medical Center of Shinshu University Hospital.