damage as early as 1880. However, reported cases have had confounding findings, such as papilledema,225,581 intraocular hypertension, and optic disc cupping,152 raising doubt as to the contribution of transsynaptic degeneration. Beatty and associates47 presented compelling histopathologic data that retrograde transsynaptic degeneration of the retinal ganglion cells with optic atrophy may occur after cerebral damage during adulthood. They presented the case of a patient who died 40 years after surgical removal of one occipital hemisphere. The vascular supply of the lateral geniculate nucleus and ipsilateral optic tract were not damaged. Using specialized staining techniques of histopathologic specimens, they demonstrated striking asymmetry of the appearance of the retinogeniculate pathway: only the lateral geniculate nucleus and optic tract on the affected side showed atrophy and axonal degeneration. This is in contradistinction to what is found in cases of optic nerve damage, where both optic tracts demonstrate atrophic axons and both lateral geniculate nuclei show atrophy in the laminae corresponding to the damaged nerve.

Band atrophy of the optic disc is most often encountered in patients with compressive lesions of the anterior visual pathway(e.g.,pituitaryadenomas,craniopharyngiomas).405,577 Concurrent damage to the optic tract or lateral geniculate body should be ruled out in patients who are suspected of having transsynaptic degeneration after acquired cerebral lesions. For example, observation of band atrophy of the contralateral optic disc in three patients with cerebral arteriovenous malformations might have been thought to represent transsynaptic degeneration across the optic tract. However, neuroimaging studies revealed abnormal deep venous drainage involving the optic tract, presumably causing direct axonal damage.333

Diagnostic and Prognostic Considerations

Although usually recognized in children with major neurologic deficits, CVI may be isolated, affecting otherwise healthy children.372 Mild variants of CVI in schoolchildren who have no other problems may be fairly common. More commonly, CVI is found in association with other neurological and systemic diseases. Associated disorders may directly arise from the same event that caused the CVI (e.g., trauma, hypoxia) or may represent a constellation of findings characteristic of a syndrome that also exhibits CVI (e.g., MELAS, meningomyelocele with hydrocephalus, X-linked adrenoleukodystrophy). Beside cerebral palsy, other associated disorders include, mental retardation, learning disabilities, seizure disorders, microcephaly, hydrocephalus, and myelomeningocele. It should be noted, however, that some children with mental retardation or autism may be mistakenly diagnosed as

having CVI because they display visual inattention, with lack of interest and detachment from their environment. The gaze avoidance that typifies autism can raise similar concerns. Although an intact visual system can often be demonstrated in such patients with the use of forced preferential looking techniques,210 this technique is not useful in children with autism.337

Improvement of vision occurs to varying degrees in most patients with CVI. In several studies of patients with CVI, over half of the children showed a significant improvement of vision on followup.218,341,613 While many patients have some recovery in visual acuity, most never see well.318 In one study, patients with the greatest improvement in visual function were those who had better initial acuity.318 When the etiology of the CVI is taken into consideration, a more accurate prediction of visual prognosis may be made. For instance, ischemic and traumatic cases are more likely to show improvement than those due to neurometabolic disorders (e.g., X-linked adrenoleukodystrophy). Seizures and microcephaly are felt to impart a worse long-term prognosis in children with cortical visual loss260a although medical control of seizures can produce striking improvement in vision. The full scope of visual recovery may, in some cases, take several years to be realized.218 However, improvement of vision that occurs after a year of the initial injury may reflect our better ability to accurately test older children and the increased ability of these children to use their limited vision with time.341 The sequence of visual recovery includes color vision, form vision and, finally, visual acuity.

Persistent visual function despite apparently severe cortical damage may be due to a variety of possibilities, such as the following: (1) Children may still have residual cortex, with some sparing of vision because central vision is widely represented in the occipital cortex. (2) Some visual recovery may be attributable to the plasticity of the brain in children, with other parts of the brain taking over via rewiring of neuronal connections, reactive synaptogenesis, rerouting of axons, and neurochemical adaptations.170 This may be interpreted by clinicians, parents, and teachers as simply learning to better “interpret” poor images. (3) Residual vision may be due to the so-called “blindsight” phenomenon. The collicular or pulvinar systems, the putative centers for blindsight, may be the area in the CNS that subserves vision even in patients with little or no cortical tissue. (4) Finally, it is theoretically possible that residual vision stems from heterotopic cortex.

Currently, the extent of cortical damage is studied with anatomical neuroimaging modalities such as CT or MRI.152 The neuroimaging abnormalities found in patients with CVI range from essentially normal to highly abnormal imaging studies, demonstrating a virtual absence of the posterior visual pathway (Fig. 1.13). Frequent findings on neuroimaging studies include diffuse cerebral atrophy, bioccipital lobe infarctions, periventricular leukomalacia, cerebral dysgenesis,

and parieto-occipital and parasagittal “watershed” infarctions. In children with hypoxic cortical insults, Lambert et al341 demonstrated a significant positive correlation between a poor visual outcome, an early age of hypoxic damage, and the degree of damage to the optic radiation. There was no statistically significant correlation between the visual outcome and the degree of damage to the striate and parastriate cortex.

Other studies have found a neuroimaging correlation with neurodevelopmental prognosis, with deep gray matter involvement on MR imaging, encephalomalacia, or periventricular leukomalacia indicating a poor prognosis,53 and normal neuroimaging indicating a favorable neurodevelopmental outcome35 Diffusion-weighted MR imaging (which identifies cytotoxic edema in acute ischemia and several other conditions)426 and diffusion tensor imaging (which depicts the three dimensional structure of the optic radiation)408 may provide valuable adjunctive information regarding mechanism of injury and anatomical integrity of the optic radiations.

Difficulties in establishing clinical-neuroimaging correlation may reflect the fact that anatomy and function are not one and the same; areas that may appear relatively spared on anatomic neuroimaging may have considerable dysfunction, and areas that appear damaged may still have persistent function. By correlating occipital lesions demonstrated on MRI with homonymous field defects, Horton and Hoyt256 demonstrated that central macular vision is more widely represented in the occipital cortex than previously thought. Therefore, even an extensive lesion of the occipital cortex is sometimes compatible with some degree of central “macular” sparing.

Functional neuroimaging studies such as positron emission tomography (PET)64 and SPECT445,525 have the added advantage of providing information regarding the functional, as opposed to anatomic, integrity of the brain on the basis of the underlying biochemistry. These studies may help delineate the site of dysfunction in cases in which anatomical neuroimaging shows little or no abnormality and vice versa. They also may enhance our understanding of the pathophysiology in cases in which results of clinical, electrophysiologic, and imaging studies appear incongruent. For instance, the clinical utility of SPECT has been documented in patients with CVI in whom MRI was either normal or inconclusive.525 The advent of functional brain imaging such as PET and SPECT scanning has shown that areas of the brain that are remote from the location of the primary insult may show concurrent impaired function, a phenomenon called diaschisis.15 This phenomenon infers that some patients with acute hemispheric injuries affecting the visual pathway may experience bilateral hemispheric symptoms through transhemispheric diaschisis. Functional neuroimaging may be particularly useful in cases with widespread nonocclusive cerebral ischemia and diffuse axonal injury from trauma in which the functional defect may be considerably greater than the anatomical lesion.526

ERG is not of clinical value in CVI, except to exclude concomitant retinal disease.183 Routine VEPs may be helpful in monitoring visual recovery, but they have limitations; they are fraught with technical and interpretational pitfalls, and their value remains controversial. Taylor and McCulloch560 have reported that flash VEPs may have a prognostic value in following young children with acute cortical blindness who have no preexisting neurologic disorders, irrespective of etiology. They demonstrated that an intact flash VEP in a previously normal child with cortical visual loss carries a favorable prognostic significance for visual recovery. Conversely, absent VEP signals carry a poor prognosis.560,561 Whiting et al613 reported that VEP mapping might be more helpful than traditional VEP recordings in the investigation of cortical blindness. In their study of 50 children with permanent CVI, the VEP map was always abnormal and showed good correlation with the CT scan results, whereas the conventional VEP recordings were abnormal in only 50% of cases. In addition to their utility as a tool to evaluate visual function, VEPs may have some value in predicting the neurodevelopmental outcome. Muttitt et al428 performed serial VEPs in a series of term infants with birth asphyxia and found good correlation between the VEPs and the neurodevelopmental outcome.

Early reports stressed the absence or marked attenuation of VEP responses in patients with acute cortical blindness, with recovery of VEP responses as vision improved over time.483 However, significant VEP signals may be recorded in some infants who are cortically blind.19 For example, Bodis-Wollner et al61 found normal VEPs to flash, pattern, and sinusoidal gratings in a blind child who had CT evidence of loss of the visual association cortex. This emphasizes the point that VEPs may be valuable in testing that the primary visual pathways are intact, but they do not test perception.176 Frank and Torres182 recorded VEPs in 30 cortically blind children as well as 30 sighted children who had a similar CNS disease. They found some degree of abnormality in all recordings but no significant difference between the two groups. In patients with neurologic disorders, flash VEPs are often abnormal even when the patient is well-sighted557 and have little prognostic value.395,396 As a rule, modest increases in visual acuity, preferential looking and VEPs occur over time.365,386 Sweep and step VEP have been shown to be a useful and repeatable way to quantitate vision in children with cortical visual loss.200,201,375,376 Highly specialized orientationreversal visual event related potentials have been found to correlate with severity of perinatal brain damage as assessed by MR imaging in preterm infants.22 However, it is not known to what extent CVI, PVL, and selective injury to higher cortical centers differentially affect the VEP.

Increased luminance causes a worsening of acuity thresholds in children with cortical visual loss.206 Vernier acuity is relatively lower than grating acuity in children with CVI.530 Because it more accurately correlates with visual acuity, and because it is cortically-mediated, Vernier acuity may provide