Stability of performance of a handheld radial shape discrimination test in patients at risk of
developing neovascular AMD
Noelia Pitrelli
1Department
Purpose
We aimed to assess the performance of the handheld Radial Shape Discrimination
(hRSD) test longitudinally in patients at risk of developing neovascular age-related
macular degeneration (nAMD) prior to disease development.
1
Vazquez ,
Simon P.
Results
73 participants performed the hRSD test on 5 consecutive visits over a period of 6.5±1.4 months (mean
interval between visits 1.6±0.5 months).
•
•
•
•
The hRSD test1
• Run on an Apple iPod Touch
• Stimuli: Radial frequency patterns
(sinusoidally modulated circles,
frequency = 8 cycles; Fig 1)
• Modulation amplitude controlled by a
staircase procedure.
• Spatial 3AFC
• Threshold expressed as a logMAR
value.
Fig 1: Example of stimuli used in
the hRSD test
Participants
Study eye:
• no nAMD or central GA
• VA ≥ 0.4 logMAR
• no development of nAMD or other
sight threatening condition over a
period of 5 clinical appointments
Fellow eye:
• nAMD, being monitored/treated
Procedures
• hRSD test, using the appropriate near correction2
• Spectralis OCT, assessed at baseline for:
• Presence/absence of at least one large druse
• Disruption of the ellipsoid zone (EZ)
Analysis
•
•
Repeated measures ANOVA for the stability of hRSD threshold over time
Regression analyses and t-tests for the effect of age, VA, large drusen, EZ
disruption and previous experience with the test on stability.
Definitions
Large drusen:
Disruption to EZ:
RPE elevation (≥ 70µm) within
1500µm of the foveal centre
Discontinuity of EZ within 1500µm of
the foveal centre (with/without drusen)
Heinrich
2
Heimann ,
Paul C.
1
Knox
of Eye and Vision Science, University of Liverpool, UK, 2St Paul’s Eye Unit, Royal Liverpool University Hospital, Liverpool, UK
Stability over time
Methods
1
Harding ,
•
•
•
•
Mean age 78±8y, range 57 to 92, 44 female
38 (52%) had at least one large druse
33 (45%) had some degree of disruption to the EZ
36 (49%) did not have any previous experience with the
hRSD test (37 had used the test at least once).
hRSD thresholds at each visit were not statistically significant different (F4,288=0.368, p=0.83; Fig 2)
Mean thresholds were consistently below the -0.37 logMAR value suggested for detection of nAMD1
The mean difference in hRSD threshold between the first and the last time points was -0.03 logMAR (Bland-Altman 95% limits of agreement: -0.42 and 0.36 logMAR; Fig 3)
The mean (±SD) slope of all individual (n=73) regression lines (test result on test visit) was -0.00014±0.000962. For an average follow up time of 189±42 days, this slope
represented an improvement in hRSD score of 0.03 logMAR over 6.5 months.
Fig 2: Mean plot
showing the
mean±95%CI for
each study visit.
Dashed line
indicates -0.37
logMAR. The mean
(±SD) hRSD
thresholds at each
time point were :
-0.53±0.17,
-0.55±0.17,
-0.55±0.15,
-0.55±0.19 and
-0.56±0.21 logMAR
Fig 3: Bland Altman
plot showing the
mean difference
(bias) and 95%
limits of
agreement for
visits 1 and 5
Discussion
Stable test performance over time prior to the development of
disease is an important aspect of any test designed to detect
the onset of nAMD and the hRSD test appears to meet this
requirement.
A number of tests are currently used to monitor patients at risk
of developing nAMD. The Amsler Grid is widely used although
studies have shown that its sensitivity for detection of nAMD is
suboptimal3. The Preferential Hyperacuity Perimeter (PHP),
which can be used either in clinic or at home, has shown
promise although there is a proportion of patients who struggle
to produce reliable results on it4.
None of our participants reported any difficulties using the
hRSD test. The test does not require a specific room luminance
and patients can do it using their usual reading glasses and an
eye patch. It can also be implemented on relatively inexpensive
handheld devices. The hRSD test is therefore very convenient
for use both in the clinic and at home.
Future work:
We are currently investigating the sensitivity and specificity of
the hRSD test for the detection of nAMD by continuing to
follow patients until a proportion convert to nAMD in the
study eye (the EDiMaD project – see below).
Conclusions
Effect of age, VA, large drusen, EZ
disruption and previous experience on test
stability
•
• the slope of individual regression lines (test result on
test visit) was not significantly correlated with age
(p=0.99) or baseline VA (p=0.43; Fig 4).
•
•
There was also no difference in mean regression slopes
between participants :
• with/without large drusen (p=0.53)
• with/without ellipsoid zone disruption (p=0.45).
• with/without previous experience with the hRSD test
(p=0.06)
Fig 5: Mean plots
showing the
mean±95%CI, for
groups of
participants
with/without large
drusen (left) and EZ
disruption (right).
Dashed line
indicates -0.37
logMAR.
hRSD test performance was stable over a period of
approximately six months in eyes at risk of developing nAMD
On average, thresholds remained consistently below the cutoff value for the hRSD test previously suggested to be
indicative of disease (-0.37 logMAR)1
No significant effects on performance of older age, worse
baseline VA or for participants with large drusen or
disruption to the EZ
References
1. Wang Y-Z et al. Handheld shape discrimination hyperacuity test on a mobile device for remote
monitoring of visual function in maculopathy. IOVS 2013;54(8):5497-505.
2. Pitrelli Vazquez N, Knox PC. A handheld radial shape discrimination hyperacuity test: assessment
of variability in a clinical population. OPO 2014;34(6):639.
Fig 4: Dot plots showing the relationship between individual participant slope (regression line of test result on test visit) and
age (left)/ baseline VA (right)
Effect of large drusen and EZ disruption on test
performance
• large drusen did not have a statistically significant effect on
hRSD performance (p=0.3)
• disruption to the EZ did have a statistically significant effect
(p=0.03) however the mean difference of 0.07 logMAR was
not clinically relevant (Fig 5).
3. Do DV et al. Detection of new-onset choroidal neovascularization using optical coherence
tomography The AMD DOC Study. Ophthalmology 2012;119(4):771-8.
4. Loewenstein A et al. Toward earlier detection of choroidal neovascularization secondary to agerelated macular degeneration. Multicenter evaluation of a preferential hyperacuity perimeter
designed as a home device. Retina-J Ret Vit Dis 2010;30(7):1058-64.
Poster pdf
Links
EDiMaD Project
Please feel free
to scan
Acknowledgements
This work was supported by a project grant from The Dunhill Medical Trust
(grant number: R283/0213). We are also grateful to Vital Art & Science Inc
for software and support and Sarah Houston for OCT grading.
[email protected]