Pattern electroretinograms and visual evoked potentials in HIV infection

Retinal disease, the most common ophthalmologic manifestation of the acquired immunodeficiency syndrome (AIDS), may be the result of opportunistic infection or may occur without evidence of intercurrent illness. ^[1] Noninfectious retinopathy, manifested by cotton-wool spots and hemorrhages on funduscopic examination, is the result of a microangiopathy that causes microinfarctions of the retinal nerve fiber layer (cotton-wool spots) and small intraretinal hemorrhages. The presence of noninfectious retinopathy on ophthalmologic examination in patients with HIV is highly associated with the stage of infection. It is seldom detected in asymptomatic individuals but is observed clinically in up to 50% of patients with AIDS, ^[1,2] and is closely associated with low CD4/CD8 ratios and impaired cognitive functions. ^[3] Postmortem examination reveals retinal cotton-wool spots in up to 75% of patients who died from AIDS. ^[1] In addition, morphometric examination of optic nerves from AIDS patients without infectious retinitis reveals axonal degeneration and approximately 40% reduction of the axonal population. ^[4] The extent and pattern of axonal loss, which shows no predilection for a particular class of axons, suggests that the changes may not only be secondary to damage at the retina, but may also reflect an AIDS-associated primary optic neuropathy.

Even though retinopathy is usually not clinically apparent until relatively late stages of HIV infection, it may be present in earlier stages and remain undetected since it generally is not accompanied by visual symptoms, ^[2] visual acuity usually remains normal, and cotton-wool spots are transient phenomena that are ophthalmoscopically visible for only a few weeks after they first appear. The pattern electroretinogram (PERG), a direct electrophysiologic measure of the functional integrity of the inner layers of the retina, is more sensitive than funduscopic examination in the detection of retinal disease in patients with non-HIV-related conditions that cause retinal microangiopathy. ^[5-7] In HIV infection, Keller et al. ^[8-10] and Mueller et al. ^[11] found reductions of PERG amplitudes not only in patients with microangiopathy, but also in those without funduscopic changes, which suggests that retinal disease is not limited to patients with demonstrable funduscopic changes. The pattern-reversal visual evoked potential (PVEP), which depends upon the functional integrity of the entire visual pathway from the eye to the primary visual cortex, is normal in early stages of HIV infection, but it is delayed in neurologically and ophthalmologically asymptomatic AIDS patients. ^[12]

The purpose of the present study is to assess retinal and optic nerve function by means of simultaneous PERG and PVEP recordings in HIV-positive subjects, with and without ophthalmoscopic evidence of retinal angiopathy, and in an HIV-negative control group, and to determine the association of electrophysiologic measures with immunologic factors, in particular with CD4⁺ T-lymphocyte counts.

Methods.

Results.

Transient and steady state PERGs and PVEPs obtained from a control and an HIV-positive subject are shown in Figure 1. The transient PERG (TPERG) exhibited an initial negative deflection (N1), followed by a larger positive wave (P1), and a subsequent negative component (N2). The transient PVEP (TPVEP) contained a positive (P100) component flanked by smaller negative waves. The steady state PERG (SSPERG) and the steady state PVEP (SSPVEP) had sinusoidal morphologies with alternating negative (N) and positive (P) components.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Pattern-reversal and steady state responses in a control subject and in an HIV-positive subject with a CD4⁺ T-lymphocyte count of 199/mL. Top traces: transient pattern-reversal responses (PERG and PVEP). Bottom traces: steady state responses (SSERG and SSVEP). Note reduced amplitudes in this HIV-positive subject under all stimulation conditions (group differences in amplitudes, however, reached statistical significance only for the TPERG P1 component).

Discussion.

Investigations in animals and in patients with ophthalmologic diseases suggest that the PERG is primarily generated in the inner retinal layers and reflects the functional integrity of ganglion cells. ^[14] Therefore, the PERG will be affected by processes that either (1) impair the input to the ganglion cells, (2) damage the ganglion cell bodies or their axons in the internal plexiform layer of the retina, or (3) compromise the optic nerve leading to retrograde ganglion cell axonal degeneration. Holder ^[15] has suggested that the TPERG contains two separable components that are differentially affected in retinal and optic nerve disease; thus, a retinopathy or maculopathy results in an amplitude reduction of the positive (P1) component with a corresponding decrease in the amplitude of the negative (N2) wave, but does not produce abnormalities confined to the N2 component. ^[15-17] In contrast, optic nerve diseases produce selective involvement of N2 or a greater reduction of the negative component than the positive wave, which may be secondary to retrograde ganglion cell degeneration. ^[15,18]

The pattern of reduction of PERG amplitude that we observed in HIV-positive patients, a selective amplitude decrease of P1, suggests that HIV infection is associated with dysfunction in the inner retinal layers. Since we did not record flash ERGs, however, we cannot rule out the possibility that degeneration in the outer retinal layers contributed to the observed TPERG abnormalities. This reduction was present even in the absence of cotton-wool spots on funduscopic examination, which suggests a higher sensitivity of the PERG to detect noninfectious, clinically silent retinopathy. Cotton-wool spots are short-lived manifestations of retinal microinfarctions, which are otherwise asymptomatic and, therefore, may have resolved prior to funduscopic examination. ^[19]

We did not find TPERG P1 latency changes in conjunction with reduced amplitudes, and neither did Holder ^[20] in a study of patients with retinal disease other than HIV. In contrast, Lorenz et al. ^[21] reported that patients with retinal diseases exhibit PERG latency delays. These discrepancies may be explained by methodologic differences among studies. We and Holder ^[20] employed higher luminance stimulation than Lorenz et al. ^[21] The use of a high-stimulus luminance may reduce the sensitivity to P1 latency delays since the P1 PERG latency saturates at high luminance levels. Luminance changes, however, have a similar parallel effect on the latencies of the PERG P1 and PVEP P100 waves that does not depend on a normal optic nerve since it also occurs in patients with optic neuritis. ^[22]

In addition to TPERG P1 amplitude reduction, patients with CD4 < 200/mL showed latency prolongation of the TPVEP P100 potential. This observation replicates the findings of an earlier study that we conducted in a larger patient population, which showed normal PVEP P100 latency in a group of asymptomatic seropositive subjects, but significantly prolonged latency in a group of neurologically asymptomatic subjects with clinically apparent AIDS. ^[12] Other investigators have also reported delayed PVEPs in asymptomatic seropositive subjects. ^[23] The latency prolongation of the cortical P100 potential in the presence of normal PERG latencies cannot be attributed to retinal disease, ^[21,24] and suggests optic nerve or retrochiasmal involvement. Thus, our findings indicate that retinal disease occurs in earlier stages of HIV infection, and central visual pathway involvement occurs only when severe immunosuppression develops.

The prolongation of the P100 PVEP latency in the CD4 < 200/mL patient group was not associated with significantly reduced PERG N2 amplitude. Such an association occurs in demyelinating, compressive and ischemic optic neuropathies, and may be secondary to retrograde ganglion cell degeneration. ^[15,18] One possible explanation for the lack of N2 amplitude changes would be that the optic radiations rather than the optic nerves are involved in HIV infection. Against this interpretation, however, are neuropathologic studies revealing axonal degeneration and loss in optic nerves from AIDS patients without infectious retinitis. ^[4] Other contributing factors may be an insufficient magnitude of retrograde ganglion cell degeneration to be detected in the PERG, small sample size, and methodologic differences from other studies, in particular stimulus contrast and luminance.

The P100 potential of the TPVEP did not demonstrate amplitude reduction. A possible explanation for the differential effect of HIV infection on PERG P1 and PVEP P100 amplitudes is that different retinal populations mediate their generation. The PVEP, largely dependent on macular function, is mediated by neural elements with receptive fields within the central few degrees of the visual field. ^[14,15] The PERG exhibits a lesser degree of macular dependence and monitors a retinal area several times larger than the foveal or macular area of stimulation required for the PVEP. ^[24-26] The relatively large size of the stimulation field employed in our study resulted in activation of the retina well beyond the macular area mediating the PVEP into the more peripheral retinal areas involved only in the generation of the PERG. This interpretation is consistent with the topography of cotton-wool spots in HIV retinopathy, which affect the entire posterior pole of the eye (central 15-degree radius) ^[2] and are not limited to the macular area.

Studies in primates suggest that the visual system is composed of two largely independent parallel subsystems, the magnocellular and parvocellular pathways, each subserved by a morphologically and physiologically distinct class of retinal ganglion cell. ^[27] Relative to the magnocellular pathway, the parvocellular pathway has more abundant retinal cells, smaller and more central receptive fields, slower conduction velocities, subserves high-resolution pattern vision and color vision, and is preferentially sensitive to high spatial frequencies (fine detail) and low temporal frequencies. In contrast, the magnocellular pathway subserves movement perception and low-resolution pattern vision and is preferentially sensitive to low spatial frequencies (coarse detail) and high temporal frequencies. Although these subsystems cannot be assessed independently in humans by means of noninvasive neurophysiologic methods, the preferential impairment of one or the other pathway can be inferred by examining response changes as the characteristics of stimuli are varied toward or away from the optimal response characteristics of the pathway. We presented stimuli at two temporal frequencies and observed that PERG amplitude reduction occurred only at low temporal frequencies. These findings suggest preferential impairment of the parvocellular pathways, and are in contrast with the preferential high temporal frequency attenuation of the PERG in patients with conditions such as Alzheimer's disease and glaucoma, ^[28-30] in which neuropathologic evidence indicates disproportionate loss of large retinal ganglion cells. Although patients with HIV infection usually do not have visual dysfunction on routine ophthalmologic examination, more sensitive psychophysical tests have demonstrated impairment of color vision and contrast sensitivity, ^[31,32] and as patients with AIDS live longer, the process of axonal loss, whether a result of multiple retinal nerve fiber layer infarcts or a primary optic nerve/retrochiasmal involvement, may progress and cause clinically relevant visual impairment.