Ophthalmic Surgery, Lasers and Imaging Retina

Clinical Science 

Hyperacuity Exam Screens for Choroidal Neovascularization in Age-Related Macular Degeneration on a Mobile Device

Jessica S. Chen, BS; Ron A. Adelman, MD, MPH, MBA, FACS

Abstract

BACKGROUND AND OBJECTIVE:

Timely treatment of age-related macular degeneration (AMD) is integral in improving outcomes. To catch choroidal neovascularization as soon as possible, patients should monitor vision at home. The objective of this study is to explore the Hyperacuity App (HAC) as a screen for progression of disease in AMD.

PATIENTS AND METHODS:

A cross-sectional, single-center study was performed with 33 subjects. Consent was obtained and patient information was protected in accordance with the protocol approved by the Yale Human Research Protection Program. A masked retinal subspecialist then graded the spectral-domain optical coherence tomography (SD-OCT) taken the same day to determine which patients required treatment. Further data about the patient were obtained through chart review.

RESULTS:

The HAC was shown to have 92.3% sensitivity and 61.5% specificity in distinguishing between patients who required treatment and those who did not require treatment.

CONCLUSION:

The HAC is a potential screen for choroidal neovascularization in AMD.

[Ophthalmic Surg Lasers Imaging Retina. 2016;47:708–715.]

Abstract

BACKGROUND AND OBJECTIVE:

Timely treatment of age-related macular degeneration (AMD) is integral in improving outcomes. To catch choroidal neovascularization as soon as possible, patients should monitor vision at home. The objective of this study is to explore the Hyperacuity App (HAC) as a screen for progression of disease in AMD.

PATIENTS AND METHODS:

A cross-sectional, single-center study was performed with 33 subjects. Consent was obtained and patient information was protected in accordance with the protocol approved by the Yale Human Research Protection Program. A masked retinal subspecialist then graded the spectral-domain optical coherence tomography (SD-OCT) taken the same day to determine which patients required treatment. Further data about the patient were obtained through chart review.

RESULTS:

The HAC was shown to have 92.3% sensitivity and 61.5% specificity in distinguishing between patients who required treatment and those who did not require treatment.

CONCLUSION:

The HAC is a potential screen for choroidal neovascularization in AMD.

[Ophthalmic Surg Lasers Imaging Retina. 2016;47:708–715.]

Introduction

Age-related macular degeneration (AMD) is the leading cause of irreversible blindness in white Americans.1 Every year, nearly 300,000 white Americans develop late-stage AMD, leading to severely impaired vision.2 Anti-vascular endothelial growth factor (VEGF) therapy is effective for treatment, improving or stabilizing vision.3,4 Timely diagnosis of choroidal neovascularization (CNV) and treatment with anti-VEGF therapy is important for the best chance of improving vision.5

Therefore, current American Academy of Ophthalmology guidelines suggest that when patients become aware of symptoms of AMD, including scotoma, metamorphopsia, or decreased vision, they should visit their ophthalmologist as soon as possible.6 The guidelines recommend patients who have a high risk for AMD or have asymptomatic AMD follow up with an ophthalmologist every 6 months to 24 months for a spectral-domain optical coherence tomography (SD-OCT) screening and slit lamp exam.6 However, continuous therapy has been shown to have better outcomes than sporadic, as-needed therapy,4 and subjects with new CNV often present with severe disease,7 suggesting that more frequent and effective home monitoring is needed in order to avoid increasing health care costs associated with more frequent visits.

The Amsler grid, in use since 1945,8 has been the standard of care for self-monitoring.9 A study of the Amsler grid suggested nearly half of scotomas were not detected, and more than half of the distortion reported corresponded to a region with a scotoma.10 Moreover, less than one in three subjects who developed CNV while using the Amsler grid sought help due to changes in the grid, with the majority presenting for other reasons.11

Several recent innovations have sought to tackle the problem, including the Preferential Hyperacuity Perimetry (PHP) device; M-Charts;12 and a handheld, shape discrimination hyperacuity test on the iPhone (Apple, Cupertino, CA).13

The Amsler grid has been shown to be a suboptimal screen for CNV based on a meta-analysis by Faes.12 Table 1 lists other innovations that have been developed. The U.S. Food and Drug Administration has approved the home-monitoring test device; the PHP; and the handheld, shape discrimination hyperacuity test.13 These devices all require physician prescription and have associated subscription costs.


Summary of Current Home-Monitoring Devices According to a Meta-Analysis by Faes12

Table 1:

Summary of Current Home-Monitoring Devices According to a Meta-Analysis by Faes12

The Hyperacuity App (HAC) is a test developed for use on an iPad (Apple, Cupertino, CA). The aim of the app is to be intuitive and easy to use for individuals without cognitive impairment up to age 90 while quantitatively assessing the individual's vision. We hypothesize that the Hyperacuity App is able to quantify the visual defects that indicate progression of CNV while remaining simple enough for the target population to use.

Patients and Methods

Subjects

Table 2 lists inclusion criteria and exclusion criteria used in the study. Thirty-six patients were recruited. Three patients were excluded from the study due to an inability to use the iPad. Of the remaining 66 eyes, 13 were excluded due to macular disease other than AMD. The demographics of the remaining eyes are seen in Table 3.


Inclusion and Exclusion Criteria

Table 2:

Inclusion and Exclusion Criteria


Subject Demographic Data Based on Classification of AMD According to the Classification Protocol

Table 3:

Subject Demographic Data Based on Classification of AMD According to the Classification Protocol

Hyperacuity App

We developed the HAC on an iOS platform for the purpose of the study. The app introduces challenges as shown in Figure 1. The distortion in the line is shown for 200 ms before the distortion becomes straight. The subject then selects on the line where he believes the distortion was. The selections are required to be within a certain perpendicular distance of the line to be considered. Audio feedback is provided for each selection to confirm that a valid selection was accomplished. After a valid selection, the screen clears to black. Subjects are then required to select the center of the empty screen for the next challenge in order to refocus the fovea on the center.


View on the screen of the Hyperacuity App during a challenge with distortion. Distortion remains on the screen for 200 ms before only the line becomes a straight dotted line. Numbers are shown at the bottom.

Figure 1.

View on the screen of the Hyperacuity App during a challenge with distortion. Distortion remains on the screen for 200 ms before only the line becomes a straight dotted line. Numbers are shown at the bottom.

Challenges are randomized during each test and can be either vertical or horizontal. Some challenges are randomized to have no distortions. Subjects are asked to select the center of the screen if no distortions are seen. A score is then generated from the data collected based on a grading system created prior to the study that was optimized to detect errors in distortion perception.

The device used for this study is an iPad 2 with the luminance set at 410 cd/m2. The black level at maximum brightness is 0.43 cd/m2.14 Distortions are distributed throughout a 700 pixel × 700 pixel distribution, which on the iPad 2 represented 5.3 inches on the screen (132 pixels per inch).

Study Design

This was a cross-sectional study performed at one center. Subjects were consented in accordance with the consent protocol approved by the Yale Human Research Protection Program. Patient information was deidentified and protected according to HIPAA requirements.

Subjects were recruited during their visit to a retinal subspecialist, where each subject was first assessed for Snellen best-corrected visual acuity (BCVA) and had intraocular pressures (IOPs) measured. The subjects were then given 2.5% Neo-Synephrine (phenylephrine hydrochloride; Bayer HealthCare, Whippany, NJ) and 1.0% Mydriacyl (tropicamide ophthalmic solution, USP; Alcon, Fort Worth, TX) in preparation for the fundus examination. At this point the subjects were then consented and participated in the study. The subjects underwent SD-OCT and were given a full dilated eye examination performed by the retinal subspecialist.

A single iPad 2 was used for all subjects, with brightness set to the same setting for each subject. The subject held the iPad at a comfortable distance — an average of 44.3 cm from the screen, which represented coverage of 4.97 mm of the macula. During the study, subjects were first trained to use the HAC and were allowed to practice until they were able to correctly complete five challenges on the app. If they were unable to complete five challenges correctly within 50 challenges despite repeated training, they were excluded from the study. Subjects took an average of 141 seconds to complete the 22 challenges included in the test. Upon completion of the HAC, they were asked about usability.

Classification Protocol

A masked retinal subspecialist read SD-OCTs with only the knowledge that all subjects were diagnosed with AMD. He assessed presence and severity of subretinal pigment epithelium (RPE) fluid, subretinal fluid, pigment epithelial defects, and RPE irregularities. Based on the SD-OCT, the researcher determined whether he would give the subject an anti-VEGF injection.

For the purposes of this study, subjects who were determined to require an anti-VEGF injection were considered “+treatment” (+tx). Subjects who did not require treatment were categorized to be either exudative AMD (−tx) or nonexudative AMD based on prior diagnosis in the electronic medical record (EMR). All patients diagnosed with exudative AMD were confirmed with fundus angiography (FA) prior to the study.

The masked ophthalmologist then reviewed all fundus photos separate from the SD-OCTs to assess for other retinal disease in the macula. These eyes were excluded.

Results

HAC Sensitivity and Specificity

Table 4 shows the contingency table and sensitivity and specificity for the HAC. Sensitivities were the same across all comparisons, whereas specificity varied depending on which groups were compared. The exudative (−tx) group had more false-positives than the nonexudative group.


Contingency Table: Sensitivity and Specificity of the HAC Test

Table 4:

Contingency Table: Sensitivity and Specificity of the HAC Test

HAC Score Versus Visual Acuity

Figure 2 shows a linear regression analysis of HAC score with visual acuity. HAC score was positively correlated with visual acuity in all groups. The exudative (+tx) group had the highest correlation between visual acuity and HAC score, whereas the nonexudative group had the lowest correlation. A Spearman analysis demonstrated that HAC score outliers did not significantly impact the correlation and demonstrated the larger variation in VA versus HAC score.


Linear regression analysis of visual acuity versus Hyperacuity App (HAC) score. Unranked analysis above, ranked analysis below. Dotted red line = exudative (+tx); dotted black line = exudative (−tx); solid blue line = nonexudative. Ranked analysis (below) shows range of ranked HAC scores around a line of perfect correlation. Pearson's and Spearman's correlation numbers are shown at the bottom.

Figure 2.

Linear regression analysis of visual acuity versus Hyperacuity App (HAC) score. Unranked analysis above, ranked analysis below. Dotted red line = exudative (+tx); dotted black line = exudative (−tx); solid blue line = nonexudative. Ranked analysis (below) shows range of ranked HAC scores around a line of perfect correlation. Pearson's and Spearman's correlation numbers are shown at the bottom.

HAC Score Versus RPE Irregularity

One-way analysis of variance showed HAC score was significantly higher in eyes with higher RPE irregularity grading (P < .0001) based on SD-OCT grading by a masked ophthalmologist.

HAC Score Versus Age

Pearson's correlation showed no linear relationship between age and HAC score in patients diagnosed with exudative AMD. In patients with nonexudative AMD, age was significantly correlated with HAC score.

Usability Survey Findings

Three of the 36 recruited patients were unable to use the device and were excluded from the study. At the conclusion of the study, subjects were asked, “Did you find the test easy to use?” Of the 33 included subjects, 30 subjects (90.9%) answered that the test was easy to use.

Discussion

The study suggests the HAC can screen for choroidal neovascularization in subjects with AMD with high sensitivity and moderate specificity.

The HAC is most specific in patients who have nonexudative AMD and least specific in patients with suppressed exudative AMD (Table 4). The lower specificity in subjects with exudative AMD is likely due to increased RPE irregularities that lead to metamorphopsia even without the presence of CNV. Eyes with a higher grading for RPE irregularities had significantly higher HAC scores (Figure 3). The difference in specificity suggests that the HAC is successful at detecting the pathological damage that persists even after leakage from exudative AMD has been treated.


Distribution of Hyperacuity App score in three grades of retinal pigment epithelium irregularity shown above with results of oneway analysis of variance below.

Figure 3.

Distribution of Hyperacuity App score in three grades of retinal pigment epithelium irregularity shown above with results of oneway analysis of variance below.

The lack of correlation between HAC score and VA was expected. A prospective, longitudinal examination of patients with AMD demonstrated that microperimetry was able to detect improvement or progression of disease in patients, whereas no changes were observed in the BCVA.15 Patients with nonexudative AMD had the lowest correlation, whereas patients with exudative AMD (+tx) had the highest correlation. The HAC is designed to detect scotoma or metamorphopsia — a patient with more active disease should have a higher HAC score and worse BCVA. In patients with nonexudative or suppressed exudative AMD, the HAC is impacted by factors other than scotoma or metamorphopsia and, therefore, there is less correlation between the score and the BCVA.

A correlation between HAC score and age is expected in patients without significant burden of disease, as older patients have a variety of reasons to perform more poorly on the app than younger patients. In patients with nonexudative AMD, age was significantly correlated (P = .007) with HAC score. However, in patients with exudative AMD (±tx), age was not significantly correlated with HAC score (Table 5), which supports that in patients with exudative AMD, the HAC score reflects factors beyond simply age-related proficiency.


Pearson's Correlation Analysis Between Age and HAC Score With Confidence Interval and P Value

Table 5:

Pearson's Correlation Analysis Between Age and HAC Score With Confidence Interval and P Value

Several limitations exist in the study. The HAC is designed as a monitoring device for patients diagnosed with exudative or nonexudative AMD. For best results, it should be used several times until the HAC score normalizes to a baseline score. At regular intervals, the patient should use the HAC to monitor for significant and consistent changes.

The study is limited by the cross-sectional nature of the study. The researchers assume that all patients have similar levels of proficiency after training on the HAC in order to compare the HAC to presence or absence of CNV. Another potential limitation is the evaluation of CNV in patients. However, a study by Wilde has shown SD-OCT to be 100% sensitive and 80.8% specific for CNV16 as compared with FA when graded by two ophthalmologists.

All patients were dilated for the study. A study has shown that dilation reduces high-contrast visual acuity by less than two letters on a Snellen chart, with no significant change in low-contrast visual acuity.17 However, no significant difference in HAC score was found between 29 eyes with an intraocular lens and 24 eyes with a natural lens (P = .51). This suggests that dilation-induced cycloplegia did not have a significant effect on HAC score.

Future studies should include a longitudinal study for patients with exudative AMD to determine whether an increase in HAC score compared to baseline correlates with a finding of CNV on SD-OCT. Further studies may also assess its use in diabetic retinopathy.

References

  1. Congdon N, O'Colmain B, Klaver CC, et al. Causes and prevalence of visual impairment among adults in the united states. Arch Ophthalmol. 2004;122(4):477–485. doi:10.1001/archopht.122.4.477 [CrossRef]
  2. Wong WL, Su X, Li X, et al. Global prevalence of age-related macular degeneration and disease burden projection for 2020 and 2040: a systematic review and meta-analysis. Lancet Glob Health. 2014;2(2):e106–e116. doi:10.1016/S2214-109X(13)70145-1 [CrossRef]
  3. Martin DF, Maguire MG, CATT Research Group et al. Ranibizumab and bevacizumab for neovascular age-related macular degeneration. N Engl J Med. 2011;364(20):1897–1908. doi:10.1056/NEJMoa1102673 [CrossRef]
  4. Peden MC, Suner IJ, Hammer ME, Grizzard WS. Long-term outcomes in eyes receiving fixed-interval dosing of anti-vascular endothelial growth factor agents for wet age-related macular degeneration. Ophthalmology. 2015;122(4):803–808. doi:10.1016/j.ophtha.2014.11.018 [CrossRef]
  5. Lim JH, Wickremasinghe SS, Xie J, et al. Delay to treatment and visual outcomes in patients treated with anti-vascular endothelial growth factor for age-related macular degeneration. Am J Ophthalmol. 2012;153(4):678–686, 686.e1–e2. doi:10.1016/j.ajo.2011.09.013 [CrossRef]
  6. Age-Related Macular Degeneration PPP - Updated 2015. http://www.aao.org/preferred-practice-pattern/age-related-macular-degeneration-ppp-2015. Published January 2015. Accessed January 1, 2016.
  7. Olsen TW, Feng X, Kasper TJ, Rath PP, Steuer ER. Fluorescein angiographic lesion type frequency in neovascular age-related macular degeneration. Ophthalmology. 2004;111(2):250–255. doi:10.1016/j.ophtha.2003.05.030 [CrossRef]
  8. Amsler M. L'Examen qualitatif de la fonction maculaire. Ophthalmologica. 1947;114(4–5):248–261. doi:10.1159/000300476 [CrossRef]
  9. Age-Related Macular Degeneration Summary Benchmark - 2015. American Academy of Ophthalmology. http://www.aao.org/summary-benchmark-detail/agerelated-macular-degeneration-summary-benchmark-. Published November 2015. Accessed January 1, 2016.
  10. Schuchard RA. Validity and interpretation of amsler grid reports. Arch Ophthalmol. 1993;111(6):776–780. doi:10.1001/archopht.1993.01090060064024 [CrossRef]
  11. Zaidi FH, Cheong-Leen R, Gair EJ, et al. The Amsler chart is of doubtful value in retinal screening for early laser therapy of subretinal membranes. The West London Survey. Eye (Lond). 2004;18(5):503–508. doi:10.1038/sj.eye.6700708 [CrossRef]
  12. Faes L, Bodmer NS, Bachmann LM, Thiel MA, Schmid MK. Diagnostic accuracy of the Amsler grid and the preferential hyperacuity perimetry in the screening of patients with age-related macular degeneration: systematic review and meta-analysis. Eye (Lond). 2014;28(7):788–796. doi:10.1038/eye.2014.104 [CrossRef]
  13. Wang YZ, He YG, Mitzel G, Zhang S, Bartlett M. Handheld shape discrimination hyperacuity test on a mobile device for remote monitoring of visual function in maculopathy. Invest Ophthalmol Vis Sci. 2013;54(8):5497–5505. doi:10.1167/iovs.13-12037 [CrossRef]
  14. Raymond SM. Apple iPad 2 LCD Display Shoot-Out. http://www.displaymate.com/iPad_2_ShootOut.htm. Published 2011. Accessed April 30, 2016.
  15. Wu Z, Ayton LN, Luu CD, Guymer RH. Longitudinal changes in microperimetry and low luminance visual acuity in age-related macular degeneration. JAMA Ophthalmol. 2015;133(4):442–448. doi:10.1001/jamaophthalmol.2014.5963 [CrossRef]
  16. Wilde C, Patel M, Lakshmanan A, et al. The diagnostic accuracy of spectral-domain optical coherence tomography for neovascular age-related macular degeneration: a comparison with fundus fluorescein angiography. Eye (Lond). 2015;29(5):602–609; quiz 610. doi:10.1038/eye.2015.44 [CrossRef]
  17. Potamitis T, Slade SV, Fitt AW, et al. The effect of pupil dilation with tropicamide on vision and driving simulator performance. Eye (Lond). 2000;14(Pt 3A):302–306. doi:10.1038/eye.2000.77 [CrossRef]

Summary of Current Home-Monitoring Devices According to a Meta-Analysis by Faes12

TestDescriptionSensitivitySpecificityCost
Amsler gridGrid on paper0.780.97Negligible
M-ChartsSeries of cards0.89100Not in market
PHPDevice for home use0.850.87$250 activation fee + $60/month
myVisionTrackApp on iPhoneUnknownUnknown$9/month subscription by prescription

Inclusion and Exclusion Criteria

Inclusion CriteriaExclusion Criteria
At least one eye with AMDOphthalmological intervention within last 24 hours
Snellen BCVA of 20/200 or betterOcclusion of sight in macula of study eye
Agreement to sign informed consentDemonstrated inability to complete exam
Ability to complete five challenges correctly within 50 challengesAny history in EMR of cognitive impairment
Visual acuity of 20/200 or worse

Subject Demographic Data Based on Classification of AMD According to the Classification Protocol

VariableDry AMDExudative AMD (−tx)Exudative AMD (+tx)Total

N = 24N = 16N = 13N = 53

Age (Years)

  Mean75.974.179.475.8
  SD9.44.85.98.1
  Range59–9567–8165–8759–95

Sex

  Male6618
  Female18101225

Race

  White22161331
  Hispanic2002

Visual Acuity (LogMAR)

  Mean0.180.450.400.32
  SD0.20.280.270.27
  Range0 – 0.98−0.04 – 10.04 – 0.84−0.04 – 1

Contingency Table: Sensitivity and Specificity of the HAC Test

Condition
Exudative (+tx)Exudative (−tx)Non-ExudativeTotals
HAC ResultPositive129627
Negative171725
13162352
Estimated Value95% Confidence Interval
Lower LimitUpper Limit
Sensitivity92.3%62.1%99.6%
Specificity A61.5%44.7%76.2%
Specificity B43.8%20.8%69.4%
Specificity C73.9%51.3%88.9%

Pearson's Correlation Analysis Between Age and HAC Score With Confidence Interval and P Value

GroupPearson's Correlation CI (95%)P Value
Non-Exudative(0.016 – 0.09).007
Exudative (−tx)(−0.129 – 0.387).301
Exudative (+tx)(−0.074 – 0.239).271
Authors

From Yale School of Medicine, New Haven, CT.

Preliminary results were presented at ARVO 2015 in Denver, CO, and read on May 4, 2015.

The authors report no relevant financial disclosures.

The authors acknowledge Sam Luo, BS, MSE, FCAS, MAAA, for data analysis and figure design for this paper.

Address correspondence to Jessica S. Chen, BS, 123 York Street, Apt. 3G, New Haven, CT 06511; email: jessica.chen@yale.edu.

Received: April 07, 2016
Accepted: May 26, 2016

10.3928/23258160-20160808-03

Sign up to receive

Journal E-contents