Visuocortical bilirubin-induced neurological dysfunction

summary

This review addresses the question whether elevated levels of total serum/plasma bilirubin (TB) cause measurable neurological effects, specifically to visuocortical functioning. Past research in the area of vision and its relation to jaundice has taken advantage of flash visual-evoked potentials (VEPs). Using a steady state VEP, we developed preliminary data suggesting that children who had jaundice with TB levels between 10 and 25 mg/dL, but who did not have kernicterus, have measurable changes in visual function, when compared to control infants who did not have jaundice. This non-invasive test offers information about vision thresholds, signal amplitudes, and suprathreshold changes after brain exposure to bilirubin. Here, we review this novel tool, the steady state VEP, and data suggesting that neurological changes occur in infants with moderately elevated TB levels.

1. Tools to measure visuocortical functioning

Why study measures of vision when there are so many aspects of cortical function that could be relevant to syndromes of bilirubin-induced neurologic dysfunction (BIND)? Vision is easily accessible, as we aim to show. No doubt there are significant differences in the manner in which vision is processed, compared with other sensory processing in other parts of the cortex, and we know that vision and its related functions can be enormously complex. Yet, it is possible that behavior of the visual cortex and associative areas could be generalizable to other parts of the neocortex. The visual cortex processes different types of visual stimuli by different mechanisms [1-3]. At the lowest levels, contrast sensitivity, grating acuity (the ability to discern thinly spaced lines), and vernier acuity (the ability to distinguish fine breaks in a line) are processed via different mechanisms, have different developmental trajectories, and may be affected differentially by different sorts of central nervous system (CNS) insults [1].

For example, in cortical visual impairment, where there is damage to the visual cortex caused by hypoxia, or metabolic derangements, vernier acuity is affected more adversely than is grating acuity [1]. The same is true for amblyopia, where one eye fails to develop normal visual acuity due to lack of use (e.g., strabismus, anisometropia).

Higher levels of vision functioning, including such disparate traits as face recognition (prosopagnosia) and visual memory, begin to blend into other cortical functions. The steady state visual-evoked potential (sVEP), for example, is capable of measuring both short and longer-term memory by presenting a subject with a stimulus referred to as an “odd-ball” paradigm [4]. In this, the infant subject is shown multiple similar faces followed by a random face. This latter stimulus will usually elicit an electroencephalographic (EEG) change that can be measured. This test probes short-term memory. Next, the infant can be shown the original “random” faces, which are no longer random as they have been seen at least once, followed by a truly random face. This test may tap longer-term memory.

A final reason for studying the visual cortex is that tests can successfully be performed on infants. Any abnormalities detected will be noted early in life, rather than having to wait to perform more complex neuropsychological tests years after exposure to bilirubin. Correlations between findings in infancy and later childhood are rare [5] and have often failed to show any subsequent problems in infants with moderately elevated total serum/plasma bilirubin (TB) levels. Nevertheless, early data on an infant's visual experience could be helpful.

Researchers have a number of tools in their armamentarium to study visuocortical and other types of neurodevelopment functioning in children. The first of these are developmental tests, such as the Bailey Test and Denver Developmental Test. There will always be a place for these tests in at-risk children, but the sensitivity of these tests is pushing researchers to find new ways to assay development through imaging, and through electrophysiology testing. Imaging and VEP tests may be only surrogates for developmental risk, but newer tools are also promising.

Whereas conventional magnetic resonance imging (MRI) is unlikely to be useful in discerning functional impairments, with the exception of advanced bilirubin toxicity, progress in neuro-imaging could play a role in identifying children at risk for BIND. Diffusion tensor imaging (DTI) has been used to study visual effects of prematurity [6] in apparently otherwise healthy infants, and this modality may have value when it comes to BIND, although the approach remains unexplored vis à vis bilirubin CNS toxicity.

Diffusion testing takes advantage of the accumulation of water in myelin disrupted by injury. Regions of interest can be selected where damage is suspected, and measures of anisotropy [6] can be made. In an experiment performed in our laboratory, we showed a correlation between diffusion parameters and grating acuity thresholds in preterm infants. Whether bilirubin has an effect on myelin deserves investigation.

The flash VEP technique involves flashing a light at an infant's eye or eyes, and measuring the visual cortex for signal amplitudes and speed of conduction. The test is inexpensive, non-invasive, and easy to interpret. Tests using this technique on jaundiced infants have shown mixed results, but with the suggestion that there may be a measurable effect on the visuocortical system [7,8].

The sweep VEP used in the experiment described below [9] measures vision responses to a stimulus that changes over a period of time. The simplest examples of vision stimuli are grating, vernier, and contrast sensitivity, where a highly visible stimulus (e.g., thick lines in grating acuity) becomes less and less visible until the infant no longer can detect it. If these lines are “tagged” by presenting them at a given temporal frequency (e.g., 3 Hz), then any statistically significant 3 Hz signal measured via electrodes placed over the visual cortex will form a tracing above background noise. The 3 Hz signal creates a signal that has amplitude, speed, and threshold (Fig. 1). When the signal can no longer be “seen,” the tracing falls into the background EEG noise. When the tracing is extrapolated to 0, the extrapolation of the line to the noise yields a vision threshold (Fig. 1). Fig. 2 shows three commonly used signals, and Fig. 3 shows the method used to present them to an infant. As a rule, this test is easiest to perform in infants (age <18 months) because the test takes about 10 min, and toddlers move, creating an increase in background noise. The test may be ideal, though, for infants aged <12 months [10,11].

Fig. 1.

Single-infant spatial frequency sweep (grating acuity) (left) and vernier offset sweeps (right). Visual-evoked potential (VEP) amplitude vs spatial frequency is a nearly linear function of spatial frequency beyond about 2–3 cycles per degree (cpd) and this part of the record is used to estimate a grating acuity of 7.35 cpd by regression to zero amplitude. A similar regression on the VEP amplitude vs vernier offset size function is used to estimate a vernier acuity of 1.88 arc min in this example. Blue bars at the top of the record indicate portions of the record with significant driven activity. The open squares indicate electroencephalogram noise levels recorded simultaneously. In a similar fashion, contrast sensitivity thresholds can be measured by showing a grating line, the contrast for which moves from very high to very low.

Fig. 2.

Three stimuli commonly used are vernier (left), contrast (center), and grating (right). The stimuli are “swept from visible to invisible, or vice versa,” with recordings as shown in Fig. 1.

Fig. 3.

In this picture, one of the authors (C.H.) is directing the infant's attention to a vernier stimulus presented on a high-resolution monitor. Over a period of about 10 s, the offsets become more and more difficult to see. The stimulus can be halted and restarted if the infant looks away.

Harmonics of the tested frequency can also be measured. Harmonic frequencies are equally spaced by the width of the fundamental frequency and can be found by repeatedly adding that frequency. For example, if the fundamental frequency (first harmonic) is 4 Hz, the frequencies of the next harmonics are: 8 Hz (second harmonic), 12 Hz (third harmonic), and 16 Hz (fourth harmonic).

In the preliminary reports described in the following sections, we measured response amplitudes and thresholds at different harmonics. This was done because some stimuli offer the best signal at a harmonic greater than the first. The higher harmonics offer somewhat different and confirming information. For example, in vernier acuity the first harmonic demonstrates vernier thresholds, whereas the second harmonic seems to convey more information about movement perception.

The sweep VEP has been used to quantify effects on the visual cortex for several pediatric diseases and toxicities. Infants with hypoxic–ischemic encephalopathy (HIE) may develop cortical visual impairment, the leading cause of low vision in the western countries [12]. Studies from our laboratory have shown negative effects in this condition, both in the acuity and signal amplitude domain [13]. In the case of toxicities, the sVEP can be used to demonstrate deleterious effects of lead exposure, formula feeding, and potentially general anesthesia. Gender differences may occur, with boys more likely to show adverse effects than girls. Premature infants (<1250 g birthweight) also show differences in their recordings compared with age-matched full-term infants. Even in low-grade intraventricular hemorrhage, we have demonstrated deleterious effects for Grade I and II hemorrhages, conditions thought to carry no developmental risk [14].

Other tests could prove of value in predicting which infant could have lasting negative CNS effects from bilirubin exposure. Untested in the BIND syndrome, but nevertheless promising, is ocular coherence tomography (OCT). Whereas this test has found its greatest use in adults with various macular and other retinal diseases, a hand-held pediatric version is available and currently is chiefly being used to study retinopathy of prematurity (ROP). The resolution of this tool is such that cell layers in the retina can be distinguished, raising the interesting question whether TB levels could be quantitated in the ganglion cell layer. Retinal tissue is considered an extension of the brain, and any correlation between OCT findings and sVEP findings, for example, could greatly simplify neonatal screening.

2. Bilirubin and the brain

Unconjugated bilirubin is a potent neurotoxin. The well-known association of high levels of TB with kernicterus seldom occurs now, thanks to measures that prevent hemolysis in blood incompatibilities, and to the advent of phototherapy and exchange transfusions. Nevertheless, there remains controversy as to what level of bilirubinemia should be treated, and which infants may be most vulnerable to the development of BIND. BIND may be permanent or transient and occurs when TB levels (a surrogate for unconjugated bilirubin in most cases) exceed an infant's neuroprotective defenses, resulting in neuronal damage in the basal ganglia, central and peripheral auditory pathways, hippocampus, diencephalon, subthalamic nuclei, midbrain, cerebellum, and the vermis [15,16]. Pontine nuclei, brainstem nuclei for ocular motor function, respiratory, neurohumoral, and electrolyte control are also affected. The spectrum of disorders may occur in vulnerable infants with exposure to bilirubin of lesser degree than is generally described [16]. Subtle neuromotor signs are associated with a range of processing disorders with objective disturbances of visuooculomotor, auditory, speech, cognition, and language among children [15,17-25]. Our preliminary data suggest that the syndrome of BIND can also affect the visual cortex and may be a permanent alteration [9].

3. Background

Our interest in bilirubin-associated vision changes was piqued a few years ago by a clinical experience. Several infants presented to one of us (W.V.G.) with a condition termed delayed visual maturation (DVM). This problem results in a transient absence of visually directed behavior, but in the presence of a completely normal eye examination, and even a normal MRI scan in most cases. DVM is now recognized to be associated with other neurological conditions, and with diseases or conditions of the anterior visual pathways. Both children examined were approximately six months of age and had a history of having had TB levels of 15–20 mg/dL.

Considerable controversy remains about the cause of DVM, and, indeed, it may be associated with or caused by a number of conditions. Even whether the abnormality is on the afferent or efferent side of the visual pathways is unknown. For example, we measured sVEPs on several infants with DVM but with no discernible bilirubin exposure and found normal VEP findings [26], very distinct from findings described below. Cocaine-exposed infants may also show DVM, but otherwise causation or association is usually unknown.

4. Preliminary data: the role of neonatal jaundice on visuocortical functioning

Intrigued by a possible association between vision function and icterus, we undertook a pilot study on infants with neonatal jaundice. Details of the study can be found in Hou et al. [9]. Here we provide a narrative description of the study, its results, and future directions.

Sixteen infants with no known hemolysis, and with TB levels between 10 and 25 mg/dL, were examined using the sVEP once they reached 12–24 weeks of age. Bilirubin measures on day 3 of life were used. Age-matched, non-jaundiced controls were used for comparison. The infants were otherwise healthy, with no condition that would affect the sVEP results. None had DVM. The following is a summary of these findings.

(1) There was a strong correlation between vernier acuity threshold and level of TB in the newborn period. The finding indicated persistent altered visuocortical altered function, lasting well beyond the period of exposure. Vernier acuity is an excellent correlate of Snellen acuity. This is a type of acuity measured in adults, easily recognized by the reader of this article as the chart that shows letters in smaller and smaller dimensions. There is reason to believe that vernier acuity also requires more cortical input than does grating or contrast sensitivity, suggesting that bilirubin may have a widespread cortical effect.

(2) Infants who had a history of neonatal jaundice showed significantly lower response amplitudes and worse or immeasurable sVEP thresholds compared to control infants for all three measures. Lower signal amplitudes have been demonstrated in conditions known to cause a reduction in cell numbers [13], such as cortical visual impairment. The finding in jaundiced infants could suggest that neuronal loss has occurred, or could be due to some other mechanism.

We measured sVEP findings in the second and fourth harmonics for grating acuity, and the first, second, and fourth harmonics for contrast and vernier acuity. We chose these harmonics because they offer the most robust signal amplitudes.

Contrast sensitivity depends partly on early visual pathway activity; i.e. the retina, but it also depends on the integrity of the retinogeniculate projections and visual cortex [9]. Elevated contrast thresholds may thus result from decreased retinal efficiency due to retinotoxic effects of hyperbilirubinemia. To a lesser degree, these early visual pathway efficiencies could have contributed to the elevations in grating and vernier acuity thresholds. It is difficult to attribute retinal toxicity for all the findings because retinal effects should affect the different response harmonics equally. We found larger response losses at the fourth harmonic for contrast and vernier functions, and this suggests that jaundice produces effects downstream from the retina, either in the geniculostriate projection or in visual cortex. The suggestion that early visual pathway effects could occur raises interest in measures of bilirubin effects on the retina alone. To date, and to our knowledge, electroretinograms have not been performed in jaundiced infants. sVEP thresholds for vernier offset were correlated with TB level, but spatial acuity and contrast sensitivity measures in the infants with neonatal jaundice were not.

5. Conclusion and future directions

These results suggest that bilirubin has an effect on visuocortical functioning at approximately three months of age, long after the infant has cleared his or her jaundice. A number of important caveats emerge from these results. The most immediate is that the effect can be construed as a negative one. Diminished (worsened) thresholds and lower signal amplitudes in groups of children suggest a diminished capacity for exposed children to process visual information. The correlation between vernier acuity and TB levels suggests an increasingly negative effect as TB levels rise for this one type of vision. Control subjects show better signal amplitudes and thresholds than exposed infants for all three types of vision.

Not answered are several important issues. First, are these findings of clinical significance? The sVEP is a very sensitive tool and may detect findings that are subclinical. Second, are these findings generalizable to the rest of the cortex? We have no way to address this question with current data, but future research could look at higher vision processing phenomena (such as visual memory, and face recognition), phenomena that are partially subserved by cortical activity outside the visual cortex. Finally, do these findings endure beyond several months of age? There is significant cortical plasticity, which could act to “dissolve” the negative effects of bilirubin exposure.

That neonatal jaundice occurs in >40% of healthy newborn children indicates that bilirubin must have positive effects on some infants. This is clearly the case, as bilirubin is a known antioxidant, and its presence eliminates or modifies the incidence of a variety of conditions, including ROP, liver, and heart disease. If visuocortical BIND is confirmed, then TB levels that should be treated will require definition, and undoubtedly there will be individual differences based on serum albumin types and quantities, and on free or unbound bilirubin levels.

Future studies should follow infants with bilirubin exposure to see whether findings persist. Given sVEP limitations, this can probably only be done to the age of about 14 months, with return visits after about age three or four years. Persistent findings will be alarming and should lead to further investigations as to clinical significance of visuocortical BIND.

Practice points.

• The sVEP offers a wide array of information on the function of the visual cortex.

• The sVEP system may be especially useful for the detection of effects of neurotoxins.

• BIND can occur in many CNS structures.

• It is still unknown which infants are the most vulnerable to the effects of bilirubin.

• Infants who exhibit altered behavior in the newborn period should have an assessment of prior bilirubin exposure.

• Delayed visual maturation, an ophthalmology problem, could be part of the spectrum of BIND.

• These data do not suggest that guidelines for bilirubin management should be changed, nevertheless they raise concerns and should inform future research.

Research directions.

• New imaging tests such as OCT could offer a means for detecting bilirubin accumulation in ganglion and other retinal cells. This would be a real-time investigation at how much bilirubin crosses the BBB.

• There is evidence that white matter may be damaged by bilirubin. DTI could be useful in measuring any effects.

• The most vulnerable infants, those born prematurely, and those with hemolysis should be studied carefully.

• Parent–infant bonding should be assessed in light of bilirubin exposure.

• Infants in the top quartile on the Bhutani nomogram should be studied for other components of BIND.

• Individual medicine will be important. Which infants require management?

• Will bilirubin-binding capacity be a useful marker for the infant who should be managed with phototherapy?