Optic Nerve Head and Nerve Layer in Alzheimer's Disease Clark S. Tsai, MD; Robert Ritch, MD; Bernard Schwartz, Neil R. Miller, MD; Thomas Chi, MD; F. Y. Hsieh, PhD \s=b\ We compared (1) the differences in the retinal nerve fiber layer between 26 patients with Alzheimer's disease and 30 age- and race-matched normal controls with use of blue-light high-resolution photography, (2) the differences in disc pallor between 30 patients with Alzheimer's disease and 32 controls with use of a boundary-tracking program and fundus photographs, and (3) the topographic disc variables between 26 patients with Alzheimer's disease and 36 controls with use of an optic nerve head analyzer. A higher proportion of patients with Alzheimer's disease had detectable nerve fiber damage as seen by red-free photography compared with controls. Although the pallor area-to-disc area ratio was not significantly different between patients with Alzheimer's disease and
lzheimer's disease (AD) is the most common form of irreversible de¬ mentia, affecting more than 7% of Americans older than 65 years.12 Clini¬ cal features include memory loss of insidious onset, aphasia, apraxias, ab¬ straction and calculation impairment, personality and behavioral changes, and visuospatial impairment.'1 Fila¬ mentous accumulations and neuronal Accepted for publication October 15, 1990. From the Departments of Ophthalmology, The New York Eye and Ear Infirmary, New York (Drs Tsai, Ritch, and Chi), Tufts-New England Medical Center, Boston, Mass (Drs Schwartz and Lee), and The Wilmer Institute, Baltimore, Md (Dr Miller); and the Department of Epidemiology, Albert Enstein College of Medicine, Bronx, NY (Dr Hsieh). Dr Tsai is now with The Kresge Eye Institute, Detroit, Mich. Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Sarasota, Fla, May 2, 1988. Reprint requests to the Glaucoma Service, The New York Eye and Ear Infirmary, New York, NY 10003 (Dr Ritch).
Fiber
MD, PhD; Shami S. Lee, MD;
controls, the patients with higher pallor area-to-disc area ratios had higher Alzheimer's
Disease Assessment Scale and longer durations of disease. Patients had an increased cupto-disc ratio and cup volume and decreased disc rim area compared with controls. These variables also correlated significantly with ADAS scores and the duration of disease. The correlation among the optic nerve head changes and the ADAS scores in patients with Alzheimer's disease suggests a potential role for optic nerve head analysis in monitoring the progression of Alzheimer's disease and in assessing the effectiveness of treatments any
(ADAS)
scores
retinal ganglion cells was found in three of four. Neither finding included neurofibrillary degeneration." Clinical¬ ly, optic disc pallor, mild to moderate optic atrophy, and/or visual field loss were found in seven of 12 patients with
AD.'"
investigated the optic disc and fiber layer of patients with AD with use of retinal nerve fiber photog¬ raphy, computer boundary tracking of the disc pallor on fundus photographs, and computerized image analysis of the optic nerve head.1"' We
nerve
PATIENTS, SUBJECTS, AND METHODS Quantitation of Impairment of Memory and Cognitive and Noncognitive Functions
developed. (Arch Ophthalmol. 1991;109:199-204)
degeneration primarily involving the
limbic system and the neocortex are found on histopathologic examination.0 The diagnosis of AD is based on clinical criteria. There is still no defi¬ nite antemortem clinical marker for AD, and brain biopsy is necessary for definitive diagnosis. Although the in¬ troduction of National Institute of Neurological and Communicative Dis¬ orders and Stroke (NINCDS) criteria has improved the diagnostic accuracy, the misdiagnosis rate remains above
10%.4·6
Neuronal
degeneration, neurofibriland neuritic plaques have discovered in the primary visual
lary tangles, been
cortex, visual association cortex, and high-order visual association cortex.'"
Pathologic evidence of optic nerve de¬ generation was found in eight of 10 postmortem specimens from patients with AD, and
a
reduced number of
Downloaded From: http://archopht.jamanetwork.com/ by a University of Manitoba User on 06/07/2015
The Alzheimer's Disease Assessment Scale (ADAS)"' includes tests for memory (word recall and word recognition), cogni¬ tion (object naming, following commands, orientation, construction praxis, and ideational praxis), and noncognitive behavior
(depression, cooperation, psychiatric
symp¬
toms, motor activity, and vegetative signs).
higher the score, the more severe the impairment. We used the ADAS to quantitate the impairment of patients with AD
The
and to exclude abnormal controls. Normal
Elderly Subjects
Individuals 55 years or older were re¬ cruited through newspaper advertising. They received medical and psychiatric eval¬ uation to exclude the disorders listed in Table 1. The remaining healthy individuals became part of a cohort of normal controls in the AD longitudinal study at the Veter¬ ans Administration Center, Bronx, NY. They underwent an extensive battery of cognitive tests semiannually and a complete medical examination annually. All had per¬ fect or near-perfect scores (0 through 10) on the ADAS. All controls recruited for this
family history of glaucoma. All patients had
Table 1.—Exclusion Criteria 1.
2. 3. 4. 5.
6. 7.
8. 9. 10.
Primary affective disorder, schizophrenia, or history of other psychosis Focal neurologic signs or symptoms History of cerebrovascular accident History of hypertension requiring antihypertensive and/or diuretic therapy Abnormal electrolyte levels or poor renal function Abnormal serum folate level, vitamin B12 level, or VDRL result Electrocardiographic abnormalities (heart rate 200 mm H20 Diagnosis of alcoholism Electroencephalographic focal findings History of head trauma with loss of consciousness for >1 h Huntingtons disease, Parkinson's disease with extrapyramidal signs and symptoms antedating memory loss, Wilson's disease, or Creutzfeldt-Jacob syndrome Findings indicative of normal-pressure hydrocephalus Uncorrectable loss of hearing or eyesight (which precludes testing)
11. Abnormal 12.
a
16. 17. 18. Active hallucinations
or
delusions
refractive
error
of less than
±3.0
diopters. Nerve Fiber
Layer Photography Nerve fiber layer photography was per¬ formed according to the technique de¬ scribed by Sommer et al.'" We used a fondus camera (Topcon Instruments, Paramus, NJ) with a power supply set at 150 W seconds, a 560-nm short-pass cutoff filter (Ditric Optics Ine, Hudson, Mass), and ex¬ tremely fine-grain and high-resolution black-and-white film (Kodak Co, Rochester, NY). Contact prints were made on perlicula
film (Kodak Co). The discs on nerve fiber layer photo¬ graphs of 26 patients with AD and 30 controls were masked with blue ink, coded, and sent to The Wilmer Institute, Balti¬ more, Md, for reading. The reader (N.R.M.) was masked regarding diagnosis. The quality of the photographs was graded
"excellent," "good," "fair," or "poor." were reported as "normal," "focal loss," "diffuse loss," "normal vs diffuse loss," or "cannot read." Only photographs graded "fair" or better were analyzed. Photographs graded "focal as
Findings
Table 2.—Patient Profiles' Sex M
F
Total
Mean ± SD Age, y
AD
15
11
NL
20
10
26 30
70.3 ± 7.6
Diagnosis Nerve fiber layer study
Pallor AD
study
NL RODA study AD
16
14
22
10
30 32
69.5 ± 7.8
15
11
26
69.0 ± 9.3
loss" or "diffuse loss" were considered defi¬ nite nerve fiber loss, while all others were considered equivocal or no nerve fiber loss (Table 3). The ADAS scores and disease duration for patients with nerve fiber layer defects were compared with those of the other patients with use of the Mann-Whit¬ ney f/test. Pallor Measurement and Statistical Analysis
NL *
AD indicates Alzheimer's disease; NL, normal; RODA, Rodenstock GMBH, Munich, Federal Republic of Germany).
optic disc analyzer (Rodenstock Instru¬
mente
Table 3.—Retinal Nerve Fiber
Grading
Quality Fair
or
Layer Photography Alzheimer's Disease 22
better Diffuse loss
or
Normal Normal
vs
Normal
plus normal
Controls 24
5*
focal loss
15 diffuse loss vs
23'
diffuse loss
Cannot read
Normal
vs
diffuse loss
26
Total Patients
Components
of 2X2 Fisher's Exact Test table.
study were white and were with the AD group (Table 2).
age-matched
Patients With AD
All patients with AD were white and underwent the same medical and psychiat¬ ric evaluation as the control group. The exclusions are listed in Table 1. All patients with AD met the NINCDS-Alzheimer's Disease-Related Disorder Association cri¬ teria for probable AD.1'
=
.075
Photographs
were taken with a Topcon with ASA 64 film and the power supply set at 36. Two clear photographs of each of 58 eyes from 30 patients with AD and 63 eyes from 32 controls (Table 2) were coded and sent to the New England Medical Center, Boston, Mass, for analysis. The reader (S.S.L.) was masked with respect to the diagnoses. The photographs were ana¬ lyzed with a computer (VAX 11/780, Digital Equipment Corp, Maynard, Mass) and boundary-tracking software capable of de¬ lineating the optic disc and area of pallor for the total disc and its four quadrants.19 This method has excellent reproducibility with a 5% coefficient of variation.I,! For 121 sets of duplicate photographs for the total disc, the mean percent coefficient of variation 5.6% ±5.9% (SD/mean 100) was (mean ± SD), with a median of 3.9% for pallor area-to-disc area (PA/DA) ratio. Each photograph was analyzed twice. Hence, each eye received four scores. These scores were averaged, and the mean was used as the final score. There was no significant difference of PA/DA ratios be¬ tween left and right eyes. Only the pallor readings of the right eye were included in statistical analyses. The Mann-Whitney U test was employed to compare the scores for statistical significance (Table 4). Spearman's correlation coefficient was used to correlate the PA/DA ratio with ADAS (cognitive, noncognitive, and total) scores and the known duration of AD (Table camera
30
(Fisher's Exact Test).
Ophthalmologic Examination All patients with AD and controls had a visual acuity of 20/25 or better, no abnormal eye examination findings, intraocular pres¬ sure less than 19 mm Hg on two separate
occasions, kinetic visual fields as deter¬ mined with the Goldmann perimeter (pa¬
tients with AD were not tested because of poor attention span), and no history of ocular trauma or surgery and/or usage of drugs or other neurotoxic agents known to cause optic atrophy. Only one control had a
Downloaded From: http://archopht.jamanetwork.com/ by a University of Manitoba User on 06/07/2015
controls and patients with AD were pared at the 1st, 2nd, 3rd, 4th, 5th,
among the operators are of a much smaller magnitude than the variability among patients. Because the RODA data included both ratio and nonratio data, the Mann-Whitney U test was used to compare the differences of disc variable between control subjects and patients with AD. To avoid a possible sampling error due to the relatively small sampling size (26 patients), Spearman's cor¬ relation coefficent analysis2'-8 was applied to correlate the RODA data obtained by one observer (R.R.) and the ADAS scores.
used in the magnification correction of the RODA measurement. The accuracy of this correction method has been demonstrated on a model eye2" and enucleated glaucoma¬
5). The percentiles of the PA/DA ratios in com¬
10th, 20th, 30th, 40th, 50th, 60th, 70th, 80th, 90th, 95th, 96th, 97th, 98th, and 99th per-
enees
monkey eyes."" Twenty-six patients with AD and
tous
36 con¬ included in this part of the study (Table 2). Seven other patients with AD were excluded due to their inability to cooperate in the RODA data collection. Only one eye of each patient was used in the analysis. In six patients with AD, only the right eye was photographed because of poor ability to concentrate. Hence, we used the right eye only. With the help of a projected stereo pair of fundus photographs, three masked experienced operators marked the disc margin separately at different times. The data were analyzed and compared by an analysis of variance model with repeated measurements between the masked opera¬ tors.2"'2. The model took three factors into consideration: diagnosis, patient, and oper¬ ator. The patient factor is nested within the diagnosis factor. The analysis was repeated for each of five variables (cup-to-disc ratio, rim area, disc area, cup volume, and rim area-to-disc area ratio). The intraclass cor¬ relation coefficients (Table 7) indicate that reliabilitv is excellent and that the differ-
centile levels with use of the StatView 512 + statistical package""' (Fig 1). The patients with AD were divided into three groups based on the total PA/DA ratio. Group 1 included all patients with a PA/DA ratio of 0.2 or less; group 2, patients with a PA/DA ratio greater than 0.2 but less than 0.4; and group 3, patients with a PA/DA ratio of 0.4 or greater but less than 0.6. We compared the ADAS (noncognitive, cognitive, and total) scores and the duration of AD among these three groups with use of the Kruskal-Wallis test (Table 6).2"
trols
Topographic Measurement of the Optic Nerve Head The technical details of the Rodenstock
optic disc analyzer (RODA; Rodenstock In¬ strumente GMBH, Munich, Federal Repub¬ lic of Germany) have been previously described,14"1 as have its reliability, reprocorrelation with clinical mea¬ ducibility, and u'2"""4 surements. Axial lengths were record¬ ed via -scan ultrasonography (Sonomed Ine, Lake Success, NY). The reading was
were
RESULTS Retinal Photographs
Photographs
were graded as "fair" better for 22 of 26 patients with AD and for 24 of 30 controls (Table 3). or
-cP
c?' 0.2-
Table 4.—Pallor Area-to-Disc Area Ratio Mean
±
0.1
SD
--
Alzheimer's Disease
Controls
P*
0.1
No. of eyes
Superior_0.236 Inferior_0.334 Nasal_0.195 Temporal_0.198 Total
pallor area-to-disc Mann-Whitney LVfest.
area
ratio
± ± ± ±
0.157_0.224 0.130_0.320 0.126_0.193 0.127_0.191
0.240 ±0.125
PA/DA Ratio
Cognitive Function
Rho
Superior
± 0.153
>.1
0.5
0.6
—
± 0.103 ± 0.120
to 5th,
10th, 20th, 30th, 40th, 50th, 60th, 70th, 80th, 90th, and 95th to 99th percentile levels.
(PA/DA) Ratio*
Total ADAS Scores Rho
Rho
0.100
0.4
Fig 1. Percentile comparison graph. Total pallor area-to-disc area (PA/DA) ratio of nor¬ mal subjects (NL) compared with PA/DA ratio of patients of Alzheimer's disease (AD) at 1st
Table 5.—ADAS Scores, Duration of Alzheimer's Disease, and Pallor Area-to-DIsc Area Function
0.3
Total Pallor Area to Disc Area Ratio in AD
± 0.108
0.230 ± 0.090
Noncognitive
0.2
Duration,
y
Rho
.1
>.1
-0.036
used. ADAS indicates Alzheimer's Disease Assessment Scale.
Table 6.—Pallor Area-to-Disc Area (PA/DA) Ratio and Status of Alzheimer's Disease* Total PA/DA Ratio PA/DA