Multifocal electroretinography in HIV-positive patients without infectious retinitis

The integrity of psychophysical visual function in non-immunocompromised PLHIV (NIPLHIV) without retinitis on antiretroviral therapy (ART)

Purpose

The present study investigated the integrity of contrast sensitivity (CS), colour vision, and pattern evoked vision potentials (VEP) in non-immunocompromised people living with HIV (NIPLHIV) without retinitis.

Methods

All participants were visually asymptomatic and no history of ocular disorders, with CD4 counts above 350 cells/mm3, low viral loads and on ART. Thirty NIPLHIV and 30 age-matched HIV negative control groups underwent F100 hue colour assessment, Pelli-Robson contrast sensitivity assessment and pattern-reversal VEP.

Results

The median F100 total error scores for NIPLHIV and controls was 33 (IQR: 28;41) and 28 (IQR: 26;48.50) respectively, this was statistically different (p= 0.020). The median P100 amplitude for NIPLHIV was 5.75 µV (IQR: 4.4;8.85) and 4.05 µV (IQR: 3.2;5.8) for controls, this was statistically different (p=0.045). The mean LogCS score 1.83±0.14 and the median P100 peak latency was 105.45 msec (IQR: 102.98;108.98) for NIPLHIV. Higher CD4+ counts were significantly associated with having higher F100 total error scores (OR=0.995; p=0.018), lower P100 amplitudes (OR=1.007; p=0.010) and higher P100 latencies (OR=0.994; p=0.011).

Conclusion

Contrast sensitivity function, colour vision, and VEP were uncompromised in NIPLHIV. Associations between CD4 counts with F100 total error scores and P100 latency may aid in the surveillance of vision of NIPLHIV.

THE THICKNESSES OF CHOROID, MACULAR SEGMENTS, PERIPAPILLARY RETINAL NERVE FIBER LAYER, AND RETINAL VASCULAR CALIBER IN HIV-1-INFECTED PATIENTS WITHOUT INFECTIOUS RETINITIS

PURPOSE

To evaluate choroidal, macular, peripapillary retinal nerve fiber layer (RNFL) thicknesses and retinal vascular caliber alterations in HIV-1-infected patients without opportunistic infections.

METHODS

This cross-sectional study included 45 HIV-1-infected patients and 47 healthy subjects. Spectral domain optical coherence tomography was used for assessment of choroidal, macular, peripapillary RNFL thicknesses and retinal vascular caliber alterations.

RESULTS

The mean CD4 count was 426 ± 226 cells per milliliter and the mean HIV-1 RNA level was 1.8 × 10 ± 3.6 × 10 copies/mL in HIV-infected group. Central inner plexiform, superior photoreceptor, superior and nasal retinal pigment epithelium layers were thinner in HIV-infected patients compared with control subjects (P < 0.05). The differences in sectoral retinal thicknesses lost their significance after Bonferroni correction (P < 0.01). The average thickness of pericentral retina within 3 mm was thinner in the photoreceptor layer in HIV-infected patients compared with control subjects (P = 0.033). The differences in peripapillary RNFL thickness, choroidal thickness, and retinal vascular caliber were not significant between the groups. Choroidal thickness and pericentral outer plexiform were thinner, whereas peripapillary RNFL was thicker in newly diagnosed cases (16 patients) compared with patients having treatment for at least 4 months or longer (27 patients, P < 0.05, Mann-Whitney U test). HIV-1 RNA showed negative correlation with choroidal thickness (r = -0.435, P = 0.003) and positive correlation with peripapillary RNFL in central (r = 0.323, P = 0.032) and superonasal (r = 0.369, P = 0.014) sectors.

CONCLUSION

Choroidal thickness was thinner in newly diagnosed patients compared with patients on treatment. Viral load showed negative correlation with choroidal thickness. Retinal segmental alterations occurred in HIV-infected patients compared with control subjects.

Meibomian gland dropout is associated with immunodeficiency at HIV diagnosis: Implications for dry eye disease

AIM

To characterize anterior eye health and tear film characteristics in individuals with human immunodeficiency virus (HIV) undergoing anti-retroviral therapy.

METHODS

This cross-sectional study involved 35 adults, categorized as healthy controls (n = 18) or as HIV-positive patients (n = 17), with no history of opportunistic infection or current ocular fundus abnormalities. Participants underwent a comprehensive anterior eye assessment. Primary outcome measures were dry eye symptoms (Ocular Surface Disease Index survey), tear film osmolarity, and extent of meibomian gland dropout. Secondary outcomes measures were ocular redness, tear film stability, and ocular surface staining. Levels of 36 cytokines were assayed from basal tears using a multiplex bead array.

RESULTS

The HIV-positive group showed more extensive meibomian gland dropout relative to controls (mean ± SD, controls: 29.6 ± 5.8 versus 37.0 ± 13.9%, p = 0.045). The extent of meibomian gland dropout was negatively correlated with blood CD4 T-cell count (a marker of immunodeficiency) at diagnosis (r = -0.69, p = 0.006). All other tests of anterior ocular health, including dry eye symptom levels, were not significantly different between the groups. There were no significant inter-group differences for the 36 cytokines assayed in the tear film.

CONCLUSIONS

We find greater meibomian gland dropout in HIV-positive individuals that is related to disease severity at diagnosis. Given this feature predisposes to dry eye disease, it suggests the need for long-term studies of anterior eye health in people with HIV.

Ophthalmologic Disease in HIV Infection: Recent Changes in Pathophysiology and Treatment

PURPOSE OF REVIEW

Ophthalmologic conditions were among the earliest described findings in patients with the acquired immunodeficiency syndrome (AIDS). The purpose of this review is to highlight recent changes in the pathophysiology and management of ophthalmologic conditions in patients infected with the human immunodeficiency virus (HIV).

RECENT FINDINGS

The introduction of highly active antiretroviral therapy (HAART) in 1996 changed ophthalmologic findings from predominantly acute infectious diseases to chronic, slowly progressive, debilitating conditions. HIV-associated neuroretinal disorder infrequently leads to blindness, but it causes visual disability in a large percentage of patients. Cytomegalovirus retinitis is now seen less commonly in the USA, but it remains an important cause of blindness in HIV-infected patients from developing countries. Immune recovery uveitis has emerged as a major cause of visual disability in the USA. As HIV has become a chronic disease, visual disability due to chronic noninfectious diseases have become increasingly important.

Pattern- and motion-related visual evoked potentials in HIV-infected adults

PURPOSE

The goal of the current study was to explore visual function in virally suppressed HIV patients undergoing combined antiretroviral therapy (cART) by using pattern-reversal and motion-onset visual evoked potentials (VEPs).

METHODS

The pattern-reversal and motion-onset VEPs were recorded in 20 adult HIV+ patients with a mean age of 38 years and CD4 cell counts ≥230 × 106 cells/L of blood.

RESULTS

Nine out of 20 patients displayed VEP abnormalities. Pattern-reversal VEPs pathology was observed in 20% of subjects, and 45% HIV patients had impaired motion-onset VEPs. Five out of 16 neurologically asymptomatic HIV patients had prolonged motion-onset VEP latencies in both eyes. Four neurologically symptomatic patients displayed simultaneously abnormal motion-onset and pattern-reversal VEP latencies: monocular involvement was observed in two patients with Lyme and cytomegalovirus unilateral optic neuritis. Binocular involvement was noted in two patients with cognitive deficits. Correlation analysis between disease duration, CD4 cell count, HIV copies in plasma, MoCA and electrophysiological parameters did not show any significant relationships.

CONCLUSIONS

The functional changes of the visual system in neurologically asymptomatic virally suppressed HIV patients displayed higher motion-onset VEP sensitivity than in standard pattern-reversal VEP examinations. This promising marker, however, has no significant association with clinical conditions. Further exploration is warranted.

Characterization of cytokine gene expression associated with noninfectious human immunodeficiency virus retinopathy in human autopsy eyes

PURPOSE

The purpose of this study was to determine the cytokine-related pathogenesis of human immunodeficiency virus retinopathy in human autopsy eyes.

METHODS

Fresh autopsy eyes were procured from clinically diagnosed patients with acquired immunodeficiency syndrome who had died as a result of disease-related complications; eyes were immediately immersed in RNAlater. Clean 2-mm trephines were used to punch individual pathologic retina in areas of cotton-wool spots and control punches. Total RNA was extracted using the TRIzol extraction protocol, and the optimal density of the RNA was measured at an optical density of 260 nm. [Delta]Ct (cytokine) values were calculated using the comparative cytokine analysis method. The results are expressed as a mean fold modulation and as a statistical comparison of Ct values controlling for retinal areas without a lesion in the same eye.

RESULTS

The fold modulations and the statistical comparisons of the cytokines studied in tissues from cotton-wool spots and control retina, respectively, regulated on activation normal T cell expressed and secreted (RANTES), macrophage inflammatory protein 1beta, macrophage inflammatory protein 1alpha (5.32x, P = 0.04), and Bcl-2-associated X protein (1.24x, P = 0.05) had a marked elevation of fold modulation and were statistically significant compared with control tissue. Interleukin-8 (1.09x, P = 0.18), interleukin-4, and interleukin-10 (2.7x, P = 0.30) were not significantly expressed in cotton-wool spots.

CONCLUSION

Certain inflammatory human immunodeficiency virus-associated and apoptotic cytokines are expressed in cotton-wool spots in eyes with human immunodeficiency virus retinopathy.

Neuroretinal Degeneration in HIV Patients Without Opportunistic Ocular Infections in the cART Era

Subtle structural and functional retinal abnormalities, termed 'HIV-associated Neuroretinal Disorder (HIV-NRD)', have been reported in HIV patients receiving combination antiretroviral therapy (cART), without infectious retinitis or any apparent fundus abnormalities otherwise. In this review, we provide an overview of studies investigating HIV-NRD in HIV patients without opportunistic ocular infections in the cART era, and try to elucidate underlying mechanisms and associated risk factors. Most studies focused on patients with severe immune-deficiency and demonstrated that patients with nadir CD4 counts<100 cells/μL are most at risk for neuroretinal damage, with a thinner retinal nerve fiber layer, subtle loss of color vision and/or contrast sensitivity, visual field deficits, and subnormal electrophysiological responses. In contrast, alterations in retinal vascular calibers and retinal blood flow were not associated with nadir CD4 counts, but instead with detectable viremia, suggesting a role for (chronic) inflammation in microvascular damage. Although the alterations in visual function are subtle, they can lead to difficulties in activities, such as reading or driving, thereby affecting quality of life. Since HIV has become a chronic disease, its long-term effects with respect to visual function loss become more important, as is recently emphasized by a longitudinal study, reporting that AIDS patients with HIV-NRD have higher risks of developing bilateral visual impairment and even blindness than patients without HIV-NRD. The question remains whether patients with high (>350 cells/μL) nadir CD4 counts and well-suppressed HIV infection on cART remain at risk for HIV-NRD, as this group constitutes a growing part of the aging HIV-infected population.

Retinal Nerve Fiber and Optic Disc Morphology in Patients with Human Immunodeficiency Virus Using the Heidelberg Retina Tomography 3

PURPOSE

To use novel confocal scanning ophthalmoscopy technology to test hypothesis that HIV-seropositive patients without history of retinitis with a history of a low CD4 count are more likely to have damage to their retinal nerve fiber layer (RNFL) when compared to patients with high CD4 count. In addition, we compared optic disc morphologic changes with glaucoma.

DESIGN

Cross-sectional study.

PARTICIPANTS AND CONTROLS

171 patients were divided into four groups. The control group consisted of 40 eyes of 20 HIV-seronegative patients. The second group consisted of 80 eyes of 41 HIV-positive patients whose CD4 cell count never dropped below 100 (1.0 x 10(9)/L). The third group consisted of 44 eyes of 26 HIV-positive patients with a history of low CD4 counts <100. Fourth group consisted of 79 eyes of 79 patients with confirmed glaucoma who served as positive controls.

TESTING

Confocal scanning laser ophthalmoscopy was performed with the Heidelberg Retina Tomograph (HRT3) and data were analyzed with HRT3, software (Heyex version 1.5.10.0).

MAIN OUTCOME MEASURES

Disc area, cup area, cup volume, rim volume, mean cup depth, maximum cup depth, cup-to-disc ration, mean RNFL thickness, and RNFL cross-sectional area.

RESULTS

Analysis of the global optic nerve and cup parameters showed no difference in disk area among the four groups. There was also no difference in cup, rim volume, mean cup depth, or maximum cup depth among the first three groups but they were all different from glaucoma group. The RNFL was thinner in glaucoma and both HIV-positive groups compared to HIV-seronegative subjects. The cross sectional RNFL area was thinner in both high and low CD4 HIV-positive groups compared to HIV-seronegative group in the nasal and temporal/inferior sectors, respectively. Glaucoma group showed thinning in all sectors.

CONCLUSIONS

HIV retinopathy results in retinal nerve fiber layer loss without structural optic nerve supportive tissue change. RNFL damage may occur early in HIV disease by mechanism different than in glaucoma.

Retinal Thickening and Photoreceptor Loss in HIV Eyes without Retinitis

Purpose

To determine the presence of structural changes in HIV retinae (i.e., photoreceptor density and retinal thickness in the macula) compared with age-matched HIV-negative controls.

Methods

Cohort of patients with known HIV under CART (combination Antiretroviral Therapy) treatment were examined with a flood-illuminated retinal AO camera to assess the cone photoreceptor mosaic and spectral-domain optical coherence tomography (SD-OCT) to assess retinal layers and retinal thickness.

Results

Twenty-four eyes of 12 patients (n = 6 HIV-positive and 6 HIV-negative) were imaged with the adaptive optics camera. In each of the regions of interest studied (nasal, temporal, superior, inferior), the HIV group had significantly less mean cone photoreceptor density compared with age-matched controls (difference range, 4,308–6,872 cones/mm²). A different subset of forty eyes of 20 patients (n = 10 HIV-positive and 10 HIV-negative) was included in the retinal thickness measurements and retinal layer segmentation with the SD-OCT. We observed significant thickening in HIV positive eyes in the total retinal thickness at the foveal center, and in each of the three horizontal B-scans (through the macular center, superior, and inferior to the fovea). We also noted that the inner retina (combined thickness from ILM through RNFL to GCL layer) was also significantly thickened in all the different locations scanned compared with HIV-negative controls.

Conclusion

Our present study shows that the cone photoreceptor density is significantly reduced in HIV retinae compared with age-matched controls. HIV retinae also have increased macular retinal thickness that may be caused by inner retinal edema secondary to retinovascular disease in HIV. The interaction of photoreceptors with the aging RPE, as well as possible low-grade ocular inflammation causing diffuse inner retinal edema, may be the key to the progressive vision changes in HIV-positive patients without overt retinitis.

Visual function assessment in simulated real-life situations in HIV-infected subjects

Visual function abnormalities are common in people living with HIV disease (PLWH) without retinitis, even after improvement in immune status. Abnormalities such as reduced contrast sensitivity, altered color vision, peripheral visual field loss, and electrophysiological changes are related to a combination of retinal dysfunctions, involving inner and outer retinal structures. The standard protocol for testing vision performance in clinical practice is the Early Treatment Diabetic Retinopathy Study (ETDRS) chart. However, this method poorly correlates with activities of daily living that require patients to assess visual stimuli in multiple light/contrast conditions, and with limited time. We utilized a novel interactive computer program (Central Vision Analyzer) to analyze vision performance in PLWH under a variety of light/contrast conditions that simulate stressful and real-world environments. The program tests vision in a time-dependent way that we believe better correlates with daily living activities than the non-timed ETDRS chart. We also aimed to correlate visual scores with retinal neuro-fiber layer thickness on optical coherence tomography. Here we show that visual acuity is more affected in PLWH in comparison to HIV-seronegative controls in varying contrast and luminance, especially if the nadir CD4+ T-cell count was lower than 100 cells/mm³. Visual impairment reflects the loss of retinal nerve fiber layer thickness especially of the temporal-inferior sector. In PLWH the ETDRS chart test led to better visual acuity compared to the Central Vision Analyzer equivalent test, likely because patients had indefinite time to guess the letters. This study confirms and strengthens the finding that visual function is affected in PLWH even in absence of retinitis, since we found that the HIV serostatus is the best predictor of visual loss. The Central Vision Analyzer may be useful in the diagnosis of subclinical HIV-associated visual loss in multiple light/contrast conditions, and may offer better understanding of this entity called “neuroretinal disorder”.

A degenerative retinal process in HIV-associated non-infectious retinopathy

HIV retinopathy is the most common non-infectious complication in the eyes of HIV-positive individuals. Oncotic lesions in the retinal nerve fiber layer, referred to as cotton wool spots (CWS), and intraretinal (IR) hemorrhages are frequently observed but are not unique to this pathology. HIV-positive patients have impaired color vision and contrast sensitivity, which worsens with age. Evidence of inner-retinal lesions and damage have been documented ophthalmoscopically, however their long term structural effect has not been investigated. It has been hypothesized that they may be partially responsible for loss of visual function and visual field. In this study we utilized clinical data, retinal imaging and transcriptomics approaches to comprehensively interrogate non-infectious HIV retinopathy. The methods employed encompassed clinical examinations, fundus photography, indirect ophthalmoscopy, Farmsworth-Munsell 100 hue discrimination testing and Illumina BeadChip analyses. Here we show that changes in the outer retina, specifically in the retinal pigment epithelium (RPE) and photoreceptor outer segments (POS) contribute to vision changes in non-infectious HIV retinopathy. We find that in HIV-positive retinae there is an induction of rhodopsin and other transcripts (including PDE6A, PDE6B, PDE6G, CNGA1, CNGB1, CRX, NRL) involved in visual transduction, as well as structural components of the rod photoreceptors (ABCA4 and ROM1). This is consistent with an increased rate of renewal of rod outer segments induced via increased phagocytosis by HIV-infected RPE previously reported in culture. Cone-specific transcripts (OPN1SW, OPN1LW, PDE6C, PDE6H and GRK7) are uniformly downregulated in HIV positive retina, likely due to a partial loss of cone photoreceptors. Active cotton wool spots and intraretinal hemorrhages (IRH) may not affect photoreceptors directly and the interaction of photoreceptors with the aging RPE may be the key to the progressive vision changes in HIV-positive patients.

Human immunodeficiency virus and its effects on the visual system

During the first 15 years of the AIDS epidemic patients experienced a high incidence of blindness due to cytomegalovirus (CMV) retinitis and other severe ocular opportunistic infections. Highly active anti-retroviral therapy, introduced in 1996, dramatically decreased the incidence of CMV retinitis. Though CMV retinitis still causes 40% of vision loss in AIDS patients, other conditions such as immune reconstitution uveitis, cataracts, and a significant othercategory -which most investigators believe is directly due to HIV - comprise the majority of cases. HIV causes vascular abnormalities of the conjunctiva and retina in the majority of AIDS patients, as well as retinitis, anterior and posterior uveitis and vasculitis. HIV frequently causes an optic neuropathy and is responsible for the majority of eye movement disorders among HIV patients. Physicians need to be aware that these problems may be the initial manifestation of HIV infections or a sign of highly active anti-retroviral therapy (HAART) failure. Therefore, patients with identifiable risk factors for AIDS who present with ophthalmologic conditions of unknown etiology should be considered for HIV testing. Finally, anti-retroviral therapy has been reported to cause asymptomatic deposits as well as degenerative conditions of both the anterior and posterior segments of the eye.

[Visual function evaluation in human immunodeficiency virus carriers]

OBJECTIVE

To determine whether there are changes in the peripapillary nerve fibre layer, in colour vision, contrast sensitivity, dark adaptation and electroretinography changes in these patients who do not have infectious retinitis.

METHODS

We studied 52 patients without ocular pathology; the mean age was 35.88 years old.

RESULTS

We observed less thickness in all quadrants, except the nasal. The colour vision was altered in 27.77% of the patients. The contrast sensitivity test showed high frequency alterations. There was no statistically significant difference in the electroretinography test or in dark adaptation.

CONCLUSIONS

There are changes in the peripapillary nerve fibre layer thickness; also we found changes in colour vision, contrast sensitivity and a decreasing trend of the B wave in the electroreninogram.

Pattern recognition can detect subtle field defects in eyes of HIV individuals without retinitis under HAART

Objectives

To use machine learning classifiers (MLCs) to seek differences in visual fields (VFs) between normal eyes and eyes of HIV+ patients; to find the effect of immunodeficiency on VFs and to compare the effectiveness of MLCs to commonly-used Statpac global indices in analyzing standard automated perimetry (SAP).

Methods

The high CD4 group consisted of 70 eyes of 39 HIV-positive patients with good immune status (CD4 counts were never <100/ml). The low CD4 group had 59 eyes of 38 HIV-positive patients with CD4 cell counts <100/ml at some period of time lasting for at least 6 months. The normal group consisted of 61 eyes of 52 HIV-negative individuals. We used a Humphrey Visual Field Analyzer, SAP full threshold program 24-2, and routine settings for evaluating VFs. We trained and tested support vector machine (SVM) machine learning classifiers to distinguish fields from normal subjects and high and CD4 groups separately. Receiver operating characteristic (ROC) curves measured the discrimination of each classifier, and areas under ROC were statistically compared.

Results

Low CD4 HIV patients: with SVM, the AUROC was 0.790 ± 0.042. SVM and MD each significantly differed from chance decision, with p < .00005. High CD4 HIV patients: the SVM AUROC of 0.664 ± 0.047 and MD were each significantly better than chance (p = .041, p = .05 respectively).

Conclusions

Eyes from both low and high CD4 HIV+ patients have VFs defects indicating retinal damage. Generalized learning classifier, SVM, and a Statpac classifier, MD, are effective at detecting HIV eyes that have field defects, even when these defects are subtle.

Structural and functional assessment in HIV-infected patients using optical coherence tomography and frequency-doubling technology perimetry

PURPOSE

To assess retinal nerve fiber layer (RNFL) and macular thickness using optical coherence tomography (OCT) on HIV-infected patients without ocular manifestations and to correlate these findings with frequency-doubling technology perimetry (FDT).

DESIGN

Cross-sectional study.

METHODS

setting: Single center. study population: Seventy-three patients (146 eyes) with clinically normal examination classified in 3 groups: group A, HIV-infected patients with CD4 count <100 cells/mm(3) for at least 6 months; group B, HIV-infected patients with CD4 count >100 cells/mm(3) since diagnosis; and group C, HIV-negative control subjects. observation procedures: Fast RNFL and fast macula scan strategies on Stratus OCT and Humphrey Matrix 24-2 full-threshold program. main outcome measures: OCT RNFL and macular thicknesses and FDT indices (mean deviation [MD], pattern standard deviation [PSD], and glaucoma hemifield test [GHT]).

RESULTS

Group A had a significantly thinner average RNFL, temporal outer macula, and inferior outer macula thicknesses when compared to groups B and C (P < .05). Statistically significant differences were observed in the FDT MD between groups A and C (P = .034) and in PSD in group A compared to groups B and C (P = .011). Eyes of HIV patients with GHT and PSD results outside normal confidence limits had thinner average RNFL thickness measures than eyes with results within normal limits in the same group of patients (P < .05).

CONCLUSIONS

HIV-infected patients with low CD4 count have a significant RNFL and macular thinning. Functional loss detected by FDT is related to RNFL thinning in HIV-infected patients.

Imaging of long-term retinal damage after resolved cotton wool spots

PURPOSE

Patients infected with the human immunodeficiency virus (HIV) develop noninfectious retinopathy characterized by retinal cotton wool spots (CWS) and microvascular abnormalities. Ophthalmoscopically, CWS fade with time. We hypothesized that structural changes should be permanent and possibly visible well after ophthalmoscopic resolution. We used simultaneous spectral domain optical coherence tomography (SD-OCT)/scanning laser ophthalmoscope (SLO) to allow colocalization of the lesions and determine the extent and location of residual damage after ophthalmoscopic resolution of the lesions.

DESIGN

Retrospective, noninterventional case series.

PARTICIPANTS

Eight eyes of 7 HIV patients with 19 resolved retinal CWS.

METHODS

Nineteen retinal CWS were imaged between 2 and 16 years (median, 7.84) after the acute lesions using simultaneous SD-OCT and SLO examinations. The areas of the previous CWS were scanned by overlaying the color retinal image over the SLO image and scanning at high resolution in the horizontal plane through the resolved lesion. Each CWS lesion had a control area taken from the same eye within 2 disc diameters of the lesion. The thickness of each of the retinal layers was compared between lesions and control areas using a paired t-test with multitest correction.

MAIN OUTCOME MEASURES

Thickness of the retinal nerve fiber layer (RNFL), ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), and outer nuclear layer (ONL).

RESULTS

The greatest loss of thickness was seen in the retinal GCL with a 43% reduction in thickness. There was a statistically significant thinning of the RNFL, GCL, IPL, INL, and OPL. The median thickness differences ranged from 5 to 7 microns. This difference was highly significant. Another striking finding was the displacement of the ONL toward the retinal surface resulting in an apparent increase in thickness of the ONL by >15% (median difference, 12 microns).

CONCLUSIONS

Our data, using ultrahigh resolution and high-speed SD-OCT/SLO, show and quantify the presence of permanent retinal destruction associated with retinal CWS in HIV disease.

Analysis with support vector machine shows HIV-positive subjects without infectious retinitis have mfERG deficiencies compared to normal eyes

PURPOSE

To test the following hypotheses: (1) eyes from individuals with human immunodeficiency virus (HIV) have electrophysiologic abnormalities that manifest as multifocal electroretinogram (mfERG) abnormalities; (2) the retinal effects of HIV in immune-competent HIV individuals differ from the effects in immune-incompetent HIV individuals; (3) strong machine learning classifiers (MLCs), like support vector machine (SVM), can learn to use mfERG abnormalities in the second-order kernel (SOK) to distinguish HIV from normal eyes; and (4) the mfERG abnormalities fall into patterns that can be discerned by MLCs. We applied a supervised MLC, SVM, to determine if mfERGs in eyes from patients with HIV differ from mfERGs in HIV-negative controls.

METHODS

Ninety-nine HIV-positive patients without visible retinopathy were divided into 2 groups: (1) 59 high-CD4 individuals (H, 104 eyes), 48.5 +/- 7.7 years, whose CD4 counts were never observed below 100, and (2) 40 low-CD4 individuals (L, 61 eyes), 46.2 +/- 5.6 years, whose CD4 counts were below 100 for at least 6 months. The normal group (N, 82 eyes) had 41 age-matched HIV-negative individuals, 46.8 +/- 6.2 years. The amplitude and latency of the first positive curve (P1, hereafter referred to as a) and the first negative curve (N1, referred to as b) in the SOK of 103 hexagon patterns of the central 28 degrees of the retina were recorded from the eyes in each group. SVM was trained and tested with cross-validation to distinguish H from N and L from N. SOK was chosen as a presumed detector of inner retinal abnormalities. Classifier performance was measured with the area under the receiver operating characteristic (AUROC) curve to permit comparison of MLCs. Improvement in performance and identification of subsets of the most important features were sought with feature selection by backward elimination.

RESULTS

In general, the SOK b-parameters separated L from N and H from N better than a-parameters, and latency separated L from N and H from N better than amplitude. In the HIV groups, on average, amplitude was diminished and latency was extended. The parameter that most consistently separated L from N and H from N was b-latency. With b-latency, SVM learned to distinguish L from N (AUROC = 0.7.30 +/- 0.044, P = .001 against chance [0.500 +/- 0.051]) and H from N (0.732 +/- 0.038, P = .0001 against chance) equally well. With best-performing subsets (21 out of 103 hexagons) derived by backward elimination, SVM distinguished L from N (0.869 +/- 0.030, P < .00005 against chance) and H from N (0.859 +/- 0.029, P <.00005 against chance) better than SVM with the full set of hexagons. Mapping the top 10 hexagon locations for L vs N and H vs N produced no apparent pattern.

CONCLUSIONS

This study confirms that mfERG SOK abnormalities develop in the retina of HIV-positive individuals. The new finding of equal severity of b-latency abnormalities in the low- and high-CD4 groups indicates that good immune status under highly active antiretroviral therapy may not protect against retinal damage and, by extension, damage elsewhere. SOKs are difficult for human experts to interpret. Machine learning classifiers, such as SVM, learn from the data without human intervention, reducing the need to rely on human skills to interpret this test.