Measurement of Urinary S100B Protein Concentrations for the Early Identification of Brain Damage in Asphyxiated Full-term Infants

Background  Perinatal asphyxia is a major cause of mortality and morbidity. To date there are no reliable methods to detect which infants will develop brain damage after asphyxia insult. We investigated whether measurements of urine levels of S100B in asphyxiated full-term newborns may be a useful tool for early detection of postasphyxia brain damage.

Methods  A prospective study of 38 infants with perinatal asphyxia and 96 control subjects, recruited at 3 tertiary departments of neonatology between April 1, 1999, and July 31, 2001. Routine laboratory variables, neurologic patterns, and urine concentrations of S100B protein were determined at 4 predetermined time points (first urination and 12, 24, and 72 hours after birth). The concentrations of S100B protein in urine were measured using an immunoluminometric assay. The results were correlated with the presence or absence of neurologic abnormalities at age 12 months.

Results  S100B protein levels were significantly higher in samples collected at all monitoring times from new-borns with abnormal neurologic findings on follow-up (first urination, 1.92 ± 0.33 µg/L; 12 hours, 2.78 ± 1.71 µg/L; 24 hours, 4.75 ± 4.08 µg/L; 72 hours, 5.93 ± 1.63 µg/L) than in samples from those without (first urination, 0.24 ± 0.06 µg/L; 12 hours, 0.13 ± 0.06 µg/L; 24 hours, 0.21 ± 0.07 µg/L; 72 hours, 0.12 ± 0.04 µg/L) or from healthy infants (first urination, 0.11 ± 0.01 µg/L; 12 hours, 0.12 ± 0.03 µg/L; 24 hours, 0.12 ± 0.02 µg/L; 72 hours, 0.12 ± 0.02 µg/L) (P<.001 for all). An S100B concentration cutoff of 0.28 µg/L at first urination had a sensitivity of 100% and a specificity of 87.3% for predicting the development of abnormal neurologic findings on follow-up. The sensitivity and specificity of measurements obtained between 12 and 72 hours were up to 100% and 98.2%, respectively.

Conclusion  Longitudinal S100B protein measurements in urine soon after birth are a useful tool to identify which asphyxiated infants are at risk of long-term neurologic sequelae.

PERINATAL ASPHYXIA constitutes an important cause of neonatal mortality and morbidity in full-term newborns, and permanent neurologic disabilities have been shown in about 25% to 28% of asphyxiated infants.^1,2 Despite accurate postnatal monitoring procedures, the postasphyxia period is crucial, since brain damage may be at a subclinical stage, or its symptoms may be hidden by the effects of sedation, while radiologic assessment may still be of no utility.^3,4 Until recently, laboratory assessment was based essentially on the determination of blood pH, the measurement of uric acid and lactate, and continuous electroencephalographic recordings.^5,6

The measurement of quantitative variables such as brain-specific biochemical markers able to diagnose subclinical lesions at stages when monitoring procedures are still unable to detect brain lesions could be especially useful in the prevention and/or management of brain injury.

S100B protein is an acidic calcium-binding protein of the EF-hand family, characterized by the most common calcium-binding motif of a helix-loop-helix structure.⁷ The protein is concentrated in the nervous system, where it is mainly located in glial cells, and has also been detected in neurons. Its half-life is about 1 hour, and it is eliminated mainly by the kidneys.⁸ The concentration of S100B in cerebrospinal fluid, in peripheral blood, and in cord blood is increased as a result of brain damage in adults, infants, and fetuses.^9-13 Increased S100B protein levels in biological fluids have also been found in preterm infants developing intraventricular hemorrhage,^14,15 suggesting a role for this peptide as a brain injury marker. In this respect, repeated measurements of S100B protein in biological fluids could be useful for monitoring newborns at risk of brain damage, such as asphyxiated infants. Perinatal asphyxia activates a cascade of pathophysiologic events that can lead to brain damage involving vasoactive agents^16,17 and calcium-mediated effects.^13-15 Among biological fluids, urine is the most suitable, as repeated sampling is easy and harmless and S100B is eliminated mainly by the kidneys.⁸

The purpose of the present study was to investigate whether measurements of urine levels of S100B in asphyxiated full-term newborns may be useful for the early detection of postasphyxia brain damage.

Thirty-eight consecutive infants with perinatal asphyxia who were born in our hospitals between April 1, 1999, and July 31, 2001, were included in the study. All asphyxiated newborns were delivered by emergency cesarean section because of acute fetal distress, defined according to the American College of Obstetricians and Gynecologists as nonreassuring fetal status (bradycardia, late deceleration of the fetal heart rate, or severe and repetitive variable deceleration of the fetal heart rate, reduced beat-to-beat variability).¹⁸ Asphyxia was defined according to an Apgar score less than 3 at the fifth minute, pH less than 7.0, or base excess greater than −12 in cord blood or venous blood taken from newborns within 60 minutes of birth, or the need for positive pressure ventilation (>3 minutes).¹⁸

One hundred twenty-one healthy term newborns served as controls. Infants admitted to the study fulfilled all of the following criteria: no maternal illness, no signs of fetal distress, pH greater than 7.2 in cord blood or venous blood, and Apgar scores at 1 and 5 minutes greater than 7. At 12-month follow-up, it was possible to monitor 96 of the 121 controls: 16 declined the monitoring program and 9 were removed from the study after discharge from the hospital because of family problems (Figure 1). Of the remaining 96 subjects, 31 had been delivered by elective cesarean section and 65 vaginally.

Infants with any malformation, systemic infection, intrauterine growth retardation, or cardiac or hemolytic disease were excluded from the study. Other exclusion criteria were multiple pregnancies; congenital or perinatal infections, including chorioamnionitis; and maternal drug addiction, hypertension, and diabetes.

Newborns in the asphyxiated group received mechanical ventilation and were sedated by means of fentanyl citrate (Fentanest; Pharmacia & Upjohn International, Milan, Italy), 0.5 to 2.5 µg/kg per hour, and midazolam hydrochloride (Ipnovel; Roche SA, Fontenay-sous-Bois, France), 50 to 400 µg/kg per hour. In all of the asphyxiated newborns, cerebral ultrasound scanning and neurologic examination were performed at the time of urine collection by a single examiner who did not know the results of the urine test. Blood was drawn by means of a catheter inserted in the cubital vein at birth, to monitor clinical and laboratory variables.

Informed consent was obtained from the parents of all patients before inclusion in the study, for which approval of the local Human Investigations Committee was obtained.

Standard cerebral ultrasonography was performed at the same times as the collection of urine by a real-time ultrasound machine (Acuson 128SP5; Acuson Corporation, Mountain View, Calif) with the use of a transducer frequency emission of 3.5 MHz. In healthy control infants, cerebral ultrasound assessment was performed at discharge from the nursery.

Neurologic examination was performed at the same times as the collection of urine. Neonatal neurologic conditions were classified by a qualitative approach as described by Prechtl.¹⁹ This test has recently been reported to be a useful support in the monitoring of high-risk newborns, thanks to its simple execution.²⁰ Each infant was assigned to 1 of 3 diagnostic groups: normal, suspect, or abnormal, in accordance with the classification used by Jurgens–van der Zee et al.²¹ An infant was considered to be abnormal when one or more of the following neurologic syndromes was unequivocally present: (1) increased or decreased excitability (hyperexcitability syndrome, convulsions, apathy syndrome, or coma); (2) increased or decreased motility (hyperkinesia or hypokinesia); (3) increased or decreased tonus (hypertonia or hypotonia); (4) asymmetries (peripheral or central); (5) defects of the central nervous system; and (6) any combination of the above. When indications of the presence of a syndrome were inconclusive or if only isolated symptoms were present, eg, mild hypotonia or only a slight tremor, the case was classified as suspect.

In the asphyxiated group, the presence within the first 7 days after birth of hypoxic ischemic encephalopathy (HIE) was classified according to the criteria described by Sarnat and Sarnat.²² The HIE was defined as mild if hyperexcitability or hypotonia persisted without seizures for at least 72 hours after birth; moderate if the infant was lethargic and had hypotonia, weak primitive reflexes, and seizures; and severe if the infant suffered frequent seizures, apnea, flaccid weakness, or coma.

Neurologic outcome was assessed at the 12th month from birth, on the basis of Amiel-Tison's criteria.²³ In particular, tone and posture, resistance against passive movements (approximation of heel to ear, ie, scarf sign; measurement of angles of certain joints, such as the popliteal angle), visual pursuit, reaching and grasping, and responses to visual and acoustic stimuli were tested. The infants were scored as normal or abnormal according to the results obtained in relation to the age in months.

The infants with perinatal asphyxia were divided into those with normal (group A; n = 20) or adverse (group B; n = 18) neurologic outcome according to whether neurologic handicaps had developed at the 12-month follow-up. Neurologic abnormalities included hypertonia-hypotonia syndrome (n = 9), hemisyndrome (n = 4), and quadriparesis (n = 2). In addition, the 3 infants who died of cardiopulmonary failure at 74, 82, and 96 hours after birth were included in group B.

S100B protein levels in urine were measured at first urination and 12, 24, and 72 hours of age. In the asphyxiated infants, a catheter was inserted into the bladder for urine sampling, owing to their critical clinical conditions and the effects of sedative drugs. In the controls, urine samples for S100B measurements were performed by a standard urine collector at the indicated times and at discharge from the nursery (72 hours of age).

At each time, urine samples were collected and immediately centrifuged at 900g for 10 minutes and the supernatants were stored at −70°C. The S100 concentration was measured in all samples by immunoluminometric assay (Lia-Mat Sangtec 100; BYK-Sangtec Diagnostica GmbH, Dietzenback, Germany). According to the manufacturer's instructions, this assay distinguishes between the A1 and B subunits of the S100 protein and measures the B subunit as defined by the 3 monoclonal antibodies SMST 12, SMSK 25, and SMSK 28. The B subunit of the S100 protein is known to be predominant (80%-96%) in the human brain.^24,25 Each measurement was performed in duplicate according to the manufacturer's recommendations, and the averages were reported. As indicated by the manufacturer, the limit of detection of the assay (minimum measurable [B₀] ± 3 SD) was 0.02 µg/L, and the precision (coefficient of variation) was 5.5% or lower within assay and 10.1% or lower between assays for concentrations ranging between 0.28 and 4.17 µg/L.

S100B concentrations are expressed as mean value ± SE. Data were analyzed for statistically significant differences between groups by Kruskal-Wallis 1-way analysis of variance and Mann-Whitney test when not normally distributed. Comparison between proportions was performed with Fisher exact test. The sensitivity, specificity, and predictive value of S100B protein levels in urine as a diagnostic test were assessed by the receiver operating characteristic curve test. Statistical significance was set at P<.05.

Table 1 shows neonatal outcomes and clinical characteristics at the different monitoring times and at 12 months in the 3 groups studied (group A, asphyxiated infants without brain damage at 12-month follow-up; group B, asphyxiated infants with brain damage at 12-month follow-up; and controls). As expected, Apgar scores at 1 and 5 minutes were significantly lower, while the incidence of respiratory distress syndrome and the need for mechanical ventilation support, the incidence of HIE and of abnormal neurologic examination results, and the presence of abnormal cerebral ultrasound patterns were significantly higher in the asphyxiated groups than in controls (P<.01, for all).

Table 2 shows the biochemical characteristics at birth in the 3 study groups. As expected, pH, partial venous carbon dioxide pressure, and base excess were significantly different in asphyxiated newborns and in controls (P<.001 for all) independent of the occurrence of brain damage. Urea nitrogen level, creatinine, and the gravity of urine in the 3 groups were similar (P>.05 for all).

At 12 hours after birth, the 3 groups presented no statistically significant differences in neonatal outcome as measured by laboratory variables, except for base excess (group A, −3.7 ± 1.1; group B, −6.1 ± 2.6; controls, 1.7 ± 0.9; P<.05). No differences were found in laboratory monitoring variables among the 3 groups at any of the monitoring times. At 24 hours, cerebral ultrasound examination was negative for cerebral bleeding in all but 3 infants, who later died (2 had subependymal hemorrhage and 1 had intraventricular hemorrhage with ventricular dilation). Periventricular hyperechogenicity was observed in 12 of 38 asphyxiated infants (5 in group A and 7 in group B; P>.05). Identical cerebral echographic patterns were observed at the 72-hour point.

Severe HIE was observed in 5 asphyxiated infants (2 in group A and 3 in group B) and moderate HIE was present in 7 infants (4 in group A and 3 in group B). Twelve of 38 asphyxiated infants were classified in the suspect group at neurologic examination (in group A, 3 with hypotonia or hypertonia and 3 with hyperexcitability; in group B, 2 with hyperexcitability and 4 with hypotonia or hypertonia). Group A suspect infants showed normal neurologic outcome at 12 months of age, whereas 6 infants in group B classified as suspect were neurologically abnormal at 12 months. Neurologic abnormalities at the 12-month follow-up included hypertonia-hypotonia syndrome (n = 9), hemisyndrome (n = 4), and quadriparesis (n = 2). At this stage, results of neurologic examination were normal both in controls and in group A infants.

Urine S100B levels were detectable in all the samples measured, and no significant differences in urine collection time at different monitoring times were shown between the asphyxia and control groups (P>.05 for all).

Urine S100B levels in the asphyxiated group were significantly higher than in controls at birth (median collection time at first urination, <2 hours) and at 12, 24, 48, and 72 hours after birth (P<.05 for all) (Table 3). Overall, S100B levels at all monitoring times were significantly higher in group B than in group A or controls (P<.001 for all) (Figure 2), while no significant difference was found between group A and controls (P>.05). No significant differences in S100B concentrations were observed between asphyxiated infants when they were subgrouped according to the occurrence of HIE (P>.05 for all). However, S100B levels were significantly higher in asphyxiated infants who developed HIE than in controls (P<.05).

The 3 infants who died at 74, 82, and 96 hours after birth showed the highest S100B levels at first urination (4.0, 3.1, and 3.0 µg/L, respectively). In this regard, it should be noted that S100B concentrations remained higher at all monitoring times in group B infants than in group A or controls, whether or not the 3 infants who died in the postnatal period were included in group B (P<.01 for all).

S100B levels in the control group did not differ when corrected for the infants' mode of delivery (0.18 ± 0.11 µg/L vs 0.21 ± 0.09 µg/L; P = .87, not significant).

The predictive values of S100B protein levels at different monitoring times for adverse neurologic outcome at 12 months are shown in Table 4.

The present study demonstrates that early urine S100B concentrations at the different monitoring times were higher in the full-term asphyxiated infants who later developed brain damage than in those who did not and in controls. These findings offer a reliable indicator of brain damage at early stages when standard monitoring procedures are of no avail.

The increased concentrations of S100B in urine are not surprising, because circulating S100B, which has already been shown to be increased in newborns with adverse neurologic outcome, is eliminated by the kidneys.^14,26 Since we found no significant differences in renal function among the 3 study groups, the higher S100B levels in the asphyxiated infants with brain damage are unlikely to be due to different degrees of urine concentration. Furthermore, since S100B is absent from kidney tissue, it is reasonable to suppose that its source in the urine is the central nervous system, which contains the protein at the stages under examination.²⁷ It is interesting, in this respect, that studies on the kinetics of S100B (conducted during open-heart surgery with cardiopulmonary bypass, which represents a condition of transitory multiple-organ failure, including kidney failure) showed that the half-life of the protein is about 1 hour and suggested that transitory alterations in kidney function due to hemodynamic changes seemed not, per se, to influence the release of S100B into the urine.^8,14 One explanation for this phenomenon may lie in the low molecular weight of the protein.⁷ The possibility that part of the S100B in the urine of asphyxiated infants may also reflect general tissue injury and renal damage thus seems fairly remote.

Ultrasound scans and laboratory variables analyzed at the predetermined monitoring times were unable to predict which of the asphyxiated infants would later develop brain damage. Neurologic examination was likewise unable to predict brain damage because of the effects of sedative drugs used during the procedure. In contrast, urine S100B levels were already higher in infants with adverse neurologic outcomes than in the asphyxiated group without such an outcome or in healthy controls. Our data agree with previous observations on preterm and term infants developing cerebral bleeding, in whom S100B levels were significantly higher at an earlier stage (24-72 hours), before the appearance of any clinical or ultrasound features suggestive of hemorrhage.^14,15,26,28 An early increase in blood S100B levels has recently been found in full-term asphyxiated infants developing hypoxic-ischemic encephalopathy²⁹ and, more recently, in asphyxiated infants developing cerebral bleeding.¹⁵ These findings provide additional support for the expedience of monitoring S100B protein in the biological fluids for use as a measurable indicator of a brain lesion before routine monitoring procedures can be performed, thus enlarging the time window for preventive and neuroprotective strategies. In addition, given that the half-life of the protein is about 1 hour,⁸ this test also offers the possibility of repeated monitoring to evaluate the efficacy of therapeutic measures. These findings are both timely and relevant in view of the use of novel and risky neuroprotective strategies, such as hypothermia, to prevent reperfusion injury. Until now it has not been known for how long the window of opportunity for intervention remains open: studies on perinatal animals have shown more rapid cell destruction and suggest that the earlier the intervention after perinatal asphyxia insult, the greater the possibility of success.^30,31 The fact that urine samples are much more easily tested than blood or cerebrospinal fluid is of special interest for improving the care of critical newborns. In this respect, it should be borne in mind that anemia caused by repeated blood sampling is a common pathologic condition in high-risk newborns.³² In addition, immunoluminometric assays for the quantification in the blood of this brain constituent are rapid (2 hours), inexpensive, and simple, and can even be carried out automatically. The fact that S100B protein measurements are not influenced by laboratory manipulation procedures, such as heparin treatment, freezing, or refrigeration, constitutes an additional advantage for their use in perinatal medicine.³³ These findings are of special interest for high-risk cases in which the clinical conditions of neonates may limit the use of computed tomography or magnetic resonance imaging investigations in the early postasphyxia phases.

Urine sampling may be limited by oliguria, which can complicate the postasphyxia period, and by the effects of sedative drugs, which may hinder urine collection in the first hours. These complications were not encountered during our research: it was possible to monitor all of the infants in our study at the predetermined monitoring times, partly because of the small amount needed for S100B protein measurements (100 µL). Unfortunately, we were unable to correlate S100B measurements with continuous electroencephalographic recordings during the first 72 hours after birth because of infrastructure limitations.

Another factor to consider is the occurrence of HIE, which may be responsible for about 25% of neurologic handicaps in these subjects.⁶ The higher S100B concentrations at first urination could be partly due to the occurrence of an acute neonatal encephalopathy²⁹ and to increased intracranial pressure and cerebral edema (particularly in the white matter), which can affect blood-brain barrier permeability. However, persistently high S100B levels were found in asphyxiated infants with neurologic handicaps independent of the occurrence of HIE, and the fact that clinical and ultrasound measures were essentially the same in the 2 asphyxiated groups also argues against this hypothesis. In addition, the wide range of urinary S100B levels at different monitoring times might be ascribed to the different degrees of brain lesion (extension), as previously reported for preterm infants.^14,15,26,28

Finally, on the basis of findings in experimental models and human studies, it has been suggested that the protein acts as a cytokine with a neurotrophic effect at physiologic concentrations^7,27,34,35 but has a neurotoxic role at high concentrations.^36,37 The possibility that at least part of the S100B measured in the urine of these patients derives from this process, which may be part of the cascade of pathologic events accompanying parenchymal damage in asphyxiated infants, should be taken into consideration.

In conclusion, the present data offer a tool for the screening of asphyxiated term infants who may later develop neurologic abnormalities and provide a new perspective for improving the monitoring and care of newborns. Further studies in wider populations in which neurologic follow-up can be continued up to 2 years of age, when possible neurologic sequelae are more clearly defined, may offer additional support for our observations.

Article

Corresponding author and reprints: Diego Gazzolo, MD, PhD, Department of Neonatology, Giannina Gaslini Children's University Hospital, Via Guglielmo Oberdan 80/1, 16167 Genoa, Italy (e-mail: dgazzolo@hotmail.com).

Accepted for publication May 29, 2003.

This study was supported in part by grant 2001063843 from COFIN (Interuniversity Cofinance), 2001, Rome, Italy (Drs Gazzolo and Michetti), and grant 200203 from the "Let's Improve Prenatal Life" Foundation, Genoa, Italy (Dr Gazzolo).

S100B protein measurement in the biological fluids seems to constitute a measurable indicator of a brain lesion before routine monitoring procedures can be performed, thus enlarging the time window for preventive and neuroprotective strategies. In addition, given that the half-life of the protein is about 1 hour, this test offers the possibility of repeated monitoring to evaluate the efficacy of therapeutic measures.

The present findings offer a tool for the screening of asphyxiated term infants who may later develop neurologic abnormalities and offer a new perspective for improving the monitoring and care of newborns.