1
Department of Ophthalmology,
University of Oulu
Head: Professor P. Juhani Airaksinen, MD,
Oulu, Finland
BLUE-ON-YELLOW PERIMETRY
IN THE DIAGNOSIS OF GLAUCOMA
by
Pait Teesalu
Academic Dissertation to be presented with the assent
of the Faculty of Medicine, University of Oulu, for
public discussion in Auditorium 5 of the University
Hospital of Oulu, on November 21st, 1998, at 12 noon.
2
CONTENTS
ABSTRACT
ABBREVIATIONS
LIST OF ORIGINAL PUBLICATIONS
1
INTRODUCTION
2
REVIEW OF THE LITERATURE
2.1
BLUE-ON-YELLOW PERIMETRY
2.1.1 Color vision changes in glaucoma
2.1.2 The basis for blue-on-yellow perimetry sensitivity
2.1.3 Blue-on-yellow versus white-on-white perimetry
2.1.4 Lens yellowing and blue-on-yellow perimetry
2.1.4 Other aspects of B/Y perimetry
2.2
AUTOFLUORESCENCE OF THE CRYSTALLINE LENS
2.3
OPTIC DISC AND RETINAL NERVE FIBER LAYER IN GLAUCOMA
2.3.1 Optic disc measurements by Heidelberg Retina Tomograph.
2.3.2 Evaluation of retinal nerve fiber layer
2.4
ASSOCIATION OF FUNCTION AND STRUCTURE
3
PURPOSE OF THE PRESENT STUDY
4
SUBJECTS AND METHODS
4.1
SUBJECTS
4.2
METHODS
4.2.1 Fluorometry of the crystalline lens
4.2.2 Visual field testing
3
4.2.3 Measurements of the optic disc
4.2.4 Evaluation of the retinal nerve fiber layer
4.2.5 Data analysis
5
RESULTS
5.1
CORRECTION OF BLUE-ON-YELLOW PERIMETRY RESULTS
5.2
ASSOCIATION WITH SCANNING CONFOCAL LASER OPTIC DISC MEASUREMENTS
5.2.1 Mean deviation of total visual field
5.2.2 Mean deviation of hemifield
5.2.3 Sensitivity and specificity
5.2.4 Normal hemifield by W/W perimetry
5.3
ASSOCIATION WITH SEMIQUANTITATIVE RNFL LOSS SCORES
5.3.1 Mean deviation of B/Y visual field
5.3.2 Normal hemifield by W/W perimetry
6
DISCUSSION
6.1 Correction of B/Y perimetry results for lens yellowing and age
6.2 Association of B/Y and W/W visual field with optic disc and RNFL
6.3 Normal W/W visual field and glaucoma
6.4 B/Y visual field in patients with ocular hypertension
7
SUMMARY AND CONCLUSIONS
REFERENCES
FIGURE LEGENDS AND FIGURES
4
ABSTRACT
The clinically detectable changes of the blue-on-yellow (B/Y) visual field as well as optic
nerve head (ONH) and retinal nerve fiber layer (RNFL) may precede white-on-white
(W/W) visual field defects in the progression of glaucoma. Optical and neural sources of
short wavelength sensitivity should be separated in the assessment of the results of B/Y
perimetry. Lens autofluorescence (AF) is directly related to lens yellowing and age.
Purpose of this study was to test the applicability of lens autofluorescence measurements in
correcting B/Y perimetry results for lens yellowing, and to evaluate the relationship
between quantitative ONH, semi-quantitative RNFL and B/Y visual field test results in
normals and patients with glaucoma and ocular hypertension.
One randomly chosen eye of 40 normal subjects and 37 patients with ocular
hypertension and different stages of glaucoma was evaluated. The B/Y and W/W visual
fields were obtained with a Humphrey perimeter. Perimetry results were adjusted for the
patients’ age and for the lens transmission index (LTI) determined as the ratio between
posterior and anterior AF peaks. A total and hemifield mean deviation (MD) of visual field
was calculated as the difference between the measured and expected mean sensitivity (MS)
values, predicted by the regression model fitted in normal subjects. The optic discs were
measured using the Heidelberg Retina Tomograph. Monochromatic RNFL photographs of
32 normal subjects and 29 patients with ocular hypertension and different stages of
glaucoma were assessed in a masked fashion.
A statistically highly significant linear correlation of B/Y MS to LTI in healthy
subjects was found. The residual standard deviation of the B/Y MS with age alone was
larger than that with LTI alone in the model (3.66 dB and 3.22 dB, respectively). The B/Y
5
visual field total and hemifield MDs showed a statistically significant correlation with
respective ONH parameters such as cup shape measure (CSM), rim volume, rim area,
mean RNFL thickness and RNFL cross sectional area as well with RNFL loss scores. With
forward stepwise logistic regression analysis using B/Y hemifield data 38% of the
glaucoma patient´s “normal”(MD > -2 dB) W/W hemifields were classified abnormal.
With the CSM alone in the model 52% of the cases were classified abnormal. The B/Y
hemifield data obtained from “normal” W/W hemifields of early glaucoma patients were
well correlated with respective RNFL loss scores found to be abnormal in 84% of
hemispheres. The analysis of variance showed a statistically significant difference between
the hemifield MD values of B/Y perimetry obtained from “normal” W/W hemifields of
normal subjects and ocular hypertensive patients with zero RNFL loss score.
The interindividual variation of the lens transmission properties increases with age.
The reference level for correcting B/Y perimetry results can be determined more precisely
using fluorometry of the lens than with age alone. The B/Y perimetry MDs are well
correlated with results of ONH and RNFL assessment. B/Y visual field, ONH and RNFL
evaluation may reveal glaucomatous defects in a hemifield found to be normal on W/W
perimetry. In subjects with ocular hypertension the functional damage detected by B/Y
perimetry may precede RNFL defects on conversion to glaucoma.
Key Words: blue-on-yellow perimetry, short-wavelength automated perimetry, glaucoma,
lens yellowing, lens autofluorescence, optic nerve head, retinal nerve fiber layer, confocal
optic disc tomography.
6
ABBREVIATIONS
AF
autofluorescence
B/Y
blue-on-yellow
CA
cup area
CDR
cup/disc area ratio
CSM
cup shape measure
CV
cup volume
DA
disc area
dB
decibel
EHMS
expected mean sensitivity of hemifield
EMS
expected mean sensitivity
HMD
mean deviation of hemifield
HMS
mean sensitivity of hemifield
HRT
Heidelberg Retina Tomograph
HVC
height variation along contour line
IOP
intraocular pressure
LWS
long-wavelength-sensitive
MD
mean deviation
7
MDC
mean depth below curved surface
mRNFLt
mean retinal nerve fiber layer thickness
MS
mean sensitivity
MWS
medium-wavelength-sensitive
MxCD
maximum cup depth
LTI
lens transmission index
ONH
optic nerve head
POAG
primary open angle glaucoma
RA
rim area
RNFL
retinal nerve fiber layer
RNFLcsa
retinal nerve fiber layer cross sectional area
RV
rim volume
SD
residual standard deviation
SWS
short-wavelength-sensitive
W/W
white-on-white
8
LIST OF ORIGINAL PUBLICATIONS
This thesis is based on the following original publication which are referred to in
the text by their Roman numerals.
I
Teesalu P, Airaksinen PJ, Tuulonen A, Nieminen H, Alanko H. (1997)
Fluorometry of the crystalline lens for correcting blue-on-yellow perimetry
results. Invest Ophthalmol Vis Sci 38:697-703.
II
Teesalu P, Vihanninjoki K, Airaksinen PJ, Tuulonen A, Läärä E. (1997)
Correlation of blue-on-yellow visual fields with scanning confocal laser
optic disc measurements. Invest Ophthalmol Vis Sci 38:2452-2459.
III
Teesalu P, Vihanninjoki K, Airaksinen PJ, Tuulonen A. (1998)
Hemifield association between blue-on-yellow visual field and optic nerve
head topographic measurements. Graefe’s Arch Clin Exp Ophthalmol
236:339-345.
IV
Teesalu P, Airaksinen PJ, Tuulonen A. (1998)
Blue-on-yellow visual field and retinal nerve fiber layer in ocular
hypertension and glaucoma. Ophthalmology 105: 2077-2081.
9
1. INTRODUCTION
Glaucoma is a progressive optic neuropathy with characteristic changes of optic
nerve head, retinal nerve fiber layer (RNFL) and visual field. The anatomic loss of
neural tissue goes along with deterioration of function and any of these three may
be the first to reach the threshold of clinical recognition. However, the detectable
RNFL and optic disc abnormalities usually precede criteria referred to as typical
glaucomatous field loss in conventional perimetry (Sommer et al. 1991, Tuulonen
et al. 1993).
In glaucoma the essential pathologic process is loss of ganglion cells and
their axons. Perimetry tests the function of ganglion cells on different location of
the retina. Conventional white-on-white (W/W) perimetry evaluates the
differential light sensitivity which is not specific for any type of ganglion cells. At
each test location, light spot stimulates most of the ganglion cell receptive fields.
The overlapping of ganglion cell receptive fields provides the eye a functional
reserve, thus a loss will be detected only if all cells at this location are
unresponsive. A substantial number of ganglion cells will die before definitive
diagnosis of glaucoma on the basis of W/W visual fields can be made (Quigley et
al. 1989).
In clinical practice it is common not to treat glaucoma suspects but to watch
them closely for detection of the earliest glaucomatous changes and then
commence a treatment in order to prevent further progression of the disease.
10
Currently, impairment of W/W perimetry draws the line between suspected and
patent glaucoma. Blue-on-yellow (B/Y) perimetry, designed to test the subgroup
of ganglion cells has shown to detect glaucomatous functional loss earlier (Sample
& Weinreb 1992, Johnson et al. 1993) and could therefore improve the adequacy
of our therapeutic decisions.
The use of B/Y perimetry as a clinical diagnostic procedure has been
complicated because of increasing lens yellowing with age. The aim of this study
was to find out whether fluorometry of the crystalline lens can be used in
correcting blue-on-yellow perimetry results for lens yellowing (I) and thereafter
evaluate the associations of lens- and age-adjusted B/Y visual field with optic
nerve head (II-III) and retinal nerve fiber layer (IV) in normal individuals and in
patients with ocular hypertension and glaucoma.
11
2. REVIEW OF THE LITERATURE
2.1. BLUE-ON-YELLOW PERIMETRY
2.1.1. Color vision changes in glaucoma
Acquired dyschromatopsias in glaucoma have been described since 1883 (Drance
1981). The presence of a tritanopic defect, established with the FarnsworthMunsell 100-Hue color vision test or the Pickford Nicholson anomaloscope, is
reported by several studies (Lakowski et al. 1972, Grützner & Schleicher 1972,
Fishman et al. 1974, Foulds et al. 1974, Adams et al. 1982). Moreover, it has been
shown (Lakowski & Drance 1979, Drance et al. 1981) that poor hue
discrimination in a tritan-like pattern may predict subsequent visual field defects.
In particular, Drance et al. (1981) found that patients with ocular hypertension and
blue-yellow color deficiencies had much higher incidence of glaucomatous visual
field loss 5 years after the initial testing than the patients with ocular hypertension
and normal color vision.
Although diffuse loss of neural tissue has been found to be the most
common pattern in glaucoma (Pederson & Anderson 1980, Sommer et al. 1991,
Tuulonen & Airaksinen 1991) the anomaloscopy could not predicted localized
nerve fiber loss because the anomaloscope measures only a 1.5° field at the
macula. Airaksinen et al. (1986) found that 25% of eyes with advanced
glaucomatous visual field loss had normal color scores suggesting that in these
eyes diffuse loss was absent or insufficient to produce abnormal color scores. In
12
consequence, subsequent investigations have been designed to examine the
sensitivity of short-wavelength-sensitive (SWS) mechanisms using principles of
perimetry.
2.1.2. The basis for blue-on-yellow perimetry sensitivity
The color perimetry has been a topic of research for many years. Isolation of shortwavelength-sensitive mechanisms using visual field testing procedures was
pioneered by the work of Marre (1972), Zisman et al. (1978), King-Smith et al.
(1984), Kitahara et al. (1982), Hart et al. (1984) and Hart & Gordon (1984).
However, it was not until the two-color increment threshold technique was applied
to automated perimetry that such procedures became a feasible clinical diagnostic
tool. The two-color increment threshold method of Stiles states that the individual
color components can be isolated or revealed by selectively reducing the
sensitivity (or influence) of competing mechanism (Stiles 1959).
Blue-on-yellow perimetry is designed to test the SWS channel selectively.
The SWS channel has its maximum sensitivity at 440 nm (blue). The mediumwavelength-sensitive (MWS) and long-wavelength-sensitive (LWS) channels,
however, have the same sensitivity at this wavelength as the SWS channel with no
background. In order to test the SWS channel selectively it is necessary to
decrease the sensitivity of MWS and LWS channels, relative to the sensitivity of
the SWS channel. This is achieved by a yellow background that contains only
13
wavelengths above the 550 nm (de Jong et al. 1992). Although maximum
separation between the sensitivity of the SWS channel and the MWS and LWS
channels is reached if the brightness is 200 cd/m2 (Yeh & Smith 1989) the
difference with the background at 100 cd/m2 is not significant (Sample et al.
1996). A background luminance of 100 cd/m2 is recommended because it provides
a better dynamic range, is more comfortable for the patients and does not require
fans to cool the system (Sample et al. 1996).
It is currently believed that the parvocellular or P-cell pathways are
responsible for processing chromatic information, especially for the shortwavelength cone system (Lee et al. 1989, Merigan & Maunsell 1993). Retinal
ganglion cells for this pathway tend to have generally smaller fiber diameters and
smaller receptive field sizes. Although the histologic studies (Quigley et al. 1987
& 1988) have shown that large ganglion cells and their large axons (magnocellular
or M-cell system) are more susceptible to glaucomatous damage the early
functional deficit has been found to be rather nonspecific affecting many visual
functions that are mediated by both pathways (Johnson 1994). The earliest deficit
in the parvocellular pathway can probably be detected by blue-on-yellow
perimetry because it tests selectively the function of isolated subset of ganglion
cells. These isolated short wavelength sensitive cells have minimal overlap of their
receptive fields due to undersampling (reduced redundancy) and thereby damage
14
to such cell types is likely to be identified more readily, even if there are
proportionately greater loss of other cells.
2.1.3. Blue-on-yellow versus white-on-white perimetry
Over the past 10 years, there has been a large accumulation of evidence that
perimetry of short-wavelength-sensitive mechanisms is presently the most
sensitive indicator of early functional losses in glaucoma patients and glaucoma
suspects. The results have shown that B/Y is superior to standard achromatic
(W/W) automated perimetry for assessing early functional glaucomatous damage
(Sample et al. 1986, Hart et al. 1990, Sample & Weinreb 1990 & 1992, Johnson et
al. 1993, Sample et al. 1993), shows deficits in ocular hypertensive and glaucoma
suspects eyes up to 3 years prior to W/W visual field loss (Sample et al. 1993,
Johnson et al. 1993). In glaucomatous eyes B/Y visual field testing indicates more
extensive damage than standard visual field. The location of the B/Y visual field
defects is often the same as that later detected with W/W visual field testing
(Sample & Weinreb 1990 & 1992, Sample et al. 1993, Johnson et al. 1993).
Some studies (Johnson et al. 1993, Sample & Weinreb 1992, Sample et al.
1995) have found B/Y perimetry to be more informative even in eyes with
advanced glaucomatous damage. Hart et al. (1990), however, found B/Y perimetry
to perform better than W/W perimetry in early defects only. The B/Y perimeter
15
with a 100 cd/m2 yellow background and a Goldmann size V (~ 1.8° of visual
angle) blue (440 nm) stimulus provide ~ 1.5 log units (15 dB) of isolation (Sample
& Weinreb 1990, Johnson et al. 1988), and when the defect depth exceeds this
value the perimetric response is no longer solely mediated by the SWS pathway
(Hart et al. 1990, Lewis et al. 1993). Felius et al. (1995) found that in cases of
extensive sensitivity losses short-wavelength detection was taken over by the
achromatic luminance channel, meaning that at a later stage of the disease the
thresholds found in B/Y perimetry are in fact luminance thresholds. In addition, in
advanced glaucoma cases the W/W perimetry should be even superior because its
larger dynamic range (maximum threshold) allows to test more advanced
functional damage.
Many patients with ocular hypertension and glaucoma are older and
therefore prone to cataract development. Moss et al. (1995) evaluated patients with
different types of cataract and age-matched normal subjects using B/Y and W/W
perimetry. They found that the amount of overall sensitivity loss produced by
cataract was approximately equal for both procedures. Patients with anterior
cortical cataract showed a slightly greater amount of sensitivity loss for B/Y than
for W/W perimetry but reverse was found in patients with posterior subcapsular
cataract. Nuclear cataract produced similar sensitivity reductions for B/Y and
W/W perimetry. The phenomenon could be explained by difference in pupil size
due to different background luminance.
16
2.1.4. Lens yellowing and blue-on-yellow perimetry
One of the factors that has limited the use of B/Y perimetry as a clinical diagnostic
procedure is the yellowing of the lens with increasing age. Previous studies
(Norren 1974, Pokorny et al. 1987, Sample et al. 1988, Johnson et al. 1989) have
reported that there is a selective absorption of short-wavelength light by the ocular
media, thereby decreasing the amount of short-wavelength light that is transmitted
to the retina. The reduced transmission causes a reduction in the height of the hill
of vision. Therefore, to distinguish short-wavelength sensitivity deficits that are
related to optic nerve damage from transmission losses occurring at the lens, it is
necessary to measure the wavelength-dependent transmission properties of the
lens. In previous studies of B/Y perimetry the absorption by ocular media was
measured using the technique of Sample et al. (1989). This involved the
measurement of two scotopic thresholds of equal sensitivity to rhodopsin (i.e., 410
nm and 560 nm) at approximately 15° eccentricity in each quadrant. The
differences in scotopic sensitivity was attributed to wavelength-dependent
absorption by the ocular media. However, measuring the lens density index using
this method is not practical in many clinical settings, because it necessitates timeconsuming dark adaptation and requires about 40 minute to complete. In addition,
it is not clear how this method works in patients with damaged retina. A video
based method using a double-pass Purkinje image reflection technique (Johnson et
17
al. 1993) and a technique using the retina as reflector for a double-pass
measurement of lens density (Delori & Burns 1996) have also been reported but
for some reason not used on B/Y perimetry procedure so far.
The problem of optical media has also been tried to solve by applying
statistical analyses to the field to factor out lens effects (Sample et al. 1994, Wild
et al. 1995). Examination of asymmetry in manner similar to the glaucoma
hemifield test, helps identify only localized B/Y perimetry deficits. The short
wavelength-sensitivity losses related to glaucomatous damage appear, however, to
have both a generalized and a localized component (Sample & Weinreb 1990, Hart
et al. 1990).
Studies have shown that there is both an increases in short-wavelength
transmission loss and greater variability from one individual to another in persons
older than 60 years (Occhipinti et al. 1986, Pokorny et al. 1987, Johnson et al.
1988, Siik et al. 1991, Sample et al. 1994). Johnson et al. (1988) and Wild et al.
(1995) found that correction for ocular media absorption did not reduce the
between-subject variability of the B/Y mean sensitivity. Wild et al. (1995)
suggested that any individual difference from the height of the average agematched normal uncorrected field because of ocular media absorption could be
removed using the pattern deviation approach. Such an approach ignores any
diffuse component from optical factors but as well from neural damage. In view of
the fact that a diffuse loss of nerve fibers is the most common pattern in early
18
glaucoma (Tuulonen & Airaksinen 1991) it seems reasonable to suggest that the
diffuse component of functional loss is not uncommon, and therefore the
correction of B/Y perimetry results also for the lens transmission losses is
warranted.
2.1.4. Other aspects of B/Y perimetry.
Humans show various asymmetries in visual sensitivity and performance between
the inferior and superior visual field. Sample et al. (1997) found significantly
greater sensitivity for the inferior B/Y visual field (superior retina) compared with
the superior field (inferior retina). The difference in mean B/Y threshold for 115
eyes tested out to 21 degrees was 2.45 ± 1.85 dB. In contrast, the difference for
achromatic threshold was much smaller (0.91± 0.87 dB). They also found that the
difference increased dramatically with eccentricity but did not increase
significantly with age.
The statistical properties of B/Y perimetry have also been examined in an
effort to provide a database, normal probability levels, and analysis procedures
that are analogous to those available for standard automated perimetry. Wild et al.
(1995) reported a greater between-subject variability for B/Y than for W/W
perimetry. The greater variability increased with the increase in eccentricity and
with the increase in age. They also found that regardless of patient age the B/Y
short-term fluctuation was larger than the W/W short-term fluctuation, but
19
although the difference reached statistical significance, the magnitude of the B/Y
short-term fluctuation remained still within the normal range encountered for the
W/W short-term fluctuation. Sample et al. (1993) reported that B/Y visual fields
have similar test-retest reliability and long-term fluctuation to that found for W/W
visual fields. The slightly higher short-term fluctuation was found also in this
study but the difference did not reach statistical significance.
The blue-cone sensitivity can also be affected by the macular pigment,
which selectively absorbs short-wavelength light. It is, however, unlikely that it
would have a meaningful influence on B/Y visual field testing because the
macular pigment extends only a few degrees beyond the fovea. Wild & Hudson
(1995) showed that the average attenuation of B/Y sensitivity at the fovea was
approximately 4 dB and declined to zero by 5.5° eccentricity.
It has been suggested that the diffuse reduction of SWS sensitivity may be
related to the degree of intraocular pressure. Lewis et al. (1993) performed B/Y
perimetry in both eyes of ocular hypertensive patients with large asymmetries in
intraocular pressure (IOP) between eyes and found that overall visual field
sensitivity was significantly depressed in the eyes with higher IOP. However,
there was no relationship between absolute IOP values and SWS sensitivity.
Finally, SWS deficits have also been recorded in low-tension glaucoma (Sample et
al. 1994).
20
2.2. AUTOFLUORESCENCE OF THE CRYSTALLINE LENS
The spectral composition of age-dependent fluorescent change in the human
crystalline lens was first measured clinically by Klang in 1948. Previous studies
(Zeimer & Noth 1984, van Best et al. 1985, Occhipinti et al. 1986, Siik et al. 1991
& 1993) indicate that autofluorescence (AF) of the lens is closely related to aging
and yellow coloration of the lens. The accumulation of lens fluorogens is believed
to be responsible to a significant degree for the increasing yellow coloration of the
lens with age (Lerman & Borkman 1976, Augusteyn 1975).
The assessment of the transmission index from the lens AF measurement is
an objective and reproducible (Siik et al. 1991), noninvasive method providing
information on the transmission properties in the central region of the lens. The
evaluation of the lens absorption by lens AF measurement was first proposed by
Zeimer & Noth (1984) and later applied and refined by van Best et al. (1985).
They assumed that the maximum AF is approximately the same in the anterior and
posterior part since the lens is rather symmetrical in structure. Any difference in
AF intensity between anterior and posterior parts can then be attributed to a loss of
exciting and fluorescent light in the lens by scatter and by absorption in the
nucleus and cortex.
The nuclear and perinuclear cortical regions of the lens show the most
intense blue-green autofluorescence (Jacobs & Krohn 1981) which may be derived
from a UV-light induced tryptophan photodegradation reaction (Lerman &
21
Borkman 1976) or by non-enzymatic glycosylation of lens proteins (Bleeker et al.
1986). However, a steady increase with age in blue and green fluorophores has
been shown only in nuclear fractions. In cortical fractions after 40 years of age,
there is no increase in blue-green fluorophores indicating that beyond this age, AF
is not a marker of aging in this region (Lerman & Borkman 1976, Yappert et al.
1992). The configurational changes of the lens proteins are the main cause of
decreased transmission in cortical cataracts (Augusteyn 1975, Lerman & Borkman
1976). Siik et al. (1991) found that the mean AF value in cortical cataracts was
even lower than in age matched healthy controls. Obviously centric cortical
opacities prevent excitation and thereby emitted light from passing through the
lens and may interfere with the measurement of nuclear AF in the blue-green
range. In eyes with cortical cataracts, therefore, the evaluation of the lens
absorption by lens AF measurement may provide erroneous results.
2.3. OPTIC DISC AND RETINAL NERVE FIBER LAYER IN GLAUCOMA
The optic nerve head and the retinal nerve fiber layer are structures in which
glaucomatous tissue damage can be detected morphologically. The essential
pathologic process in glaucoma is a loss of ganglion cell axons. As axons are lost,
the amount of neural tissue in the neural rim decreases, resulting in alterations in
the appearance of the neural rim and configuration of the optic cup. It has long
been recognized that glaucoma is associated with an increase in the absolute size
22
of the optic cup as well as the cup to disc ratio. Neither by itself is very helpful in
distinguishing the normal from the glaucomatous disc since both are the functions
of the size of the disc which varies tremendously in the normal population (Jonas
et al. 1988). Studies (Airaksinen et al. 1985, Caprioli & Miller 1988, Jonas &
Neumann 1991) have showed that the neural rim area is reduced in glaucoma. As
is true of the features of the optic cup the absolute measurement of the neural rim
area may not be very useful in diagnosing glaucoma, unless it is observed to
change with time. Airaksinen et al. (1995) showed that the rate of rim area
decrease over time was significantly greater in glaucoma and glaucoma suspect
patients than the age related decline found in normals.
In addition to optic nerve head, the atrophy of ganglion cell axons can be
observed also in the retinal nerve fiber layer. The axons are gathered together in
optic nerve head but spread out in a thin layer in the retina. Therefore, the
examination of the RNFL may even provide information of the minor losses of
axons that cannot be detected by evaluating the neuroretinal rim of the optic nerve
head. Retinal nerve fiber layer abnormalities in patients with glaucoma were first
reported by Hoyt et al in 1973. The first observable changes he found were thin,
slit-like defects or grooves in the arcuate area of the RNFL. In further progressed
cases wedge-shaped localized defects developed.
In glaucoma, two distinct patterns of neural tissue loss are recognized,
localized and diffuse. In any individual patient, one or the other pattern may
23
predominate, or both patterns may occur (Iwata et al. 1981, Airaksinen et al.
1984). It has been found that diffuse loss of axons was the most common pattern
in early glaucoma, but that a mixed pattern of diffuse and localized loss was more
common as a later finding (Tuulonen & Airaksinen 1991).
2.3.1. Optic disc measurements by Heidelberg Retina Tomograph.
Although the clinical assessment of optic nerve head has been of interest over a
century the techniques have been largely qualitative. The introduction of
computerized instruments, such as the Heidelberg Retina Tomograph, has made it
possible to obtain rapid, accurate and reproducible three-dimensional analysis of
the optic disc structure by quantitative serial point-by-point comparison of the
surface contour (Kruse et al. 1989, Weinreb et al. 1989, Dreher et al. 1991, Dreher
& Weinreb 1991, Cioffi et al. 1993, Mikelberg et al. 1993, Rohrschneider et al.
1993 & 1994, Lusky et al. 1993, Chauhan et al. 1994). Dilatation of the patient´s
pupil is not required for recording of the image data. A pupil diameter of 1 mm
was found to be sufficient to receive useful data (Lusky et al. 1993).
In addition to two-dimensional parameters such as the cup area and
neuroretinal rim area also three-dimensional parameters can be measured. For
three-dimensional measurements of the disc, the margin of the ONH has to be
defined by the operator. The calculation of the morphometric parameters is not
influenced by the operator, the HRT automatically computes data for the entire
24
nerve head on advise. In addition, the operator has the possibility to select a
segment of interest and to compute the corresponding data for this part of the
ONH on request (Burk et al. 1990).
Some three-dimensional parameters such as the maximum cup depth and
cup shape measure have the advantage of being independent of optic disc
reference levels. Others, like rim volume, cup volume, mean RNFL thickness,
describing either volume or distance are dependent of the topographic reference
plane. The selection of the reference plane has been a major issue for discussion.
In the HRT Operation Software Release 1.08, 1.09 and 1.10 a reference plane at a
location 320 m posteriorly of the mean peripapillary retinal surface height was
used as default. However, with the progression of the glaucomatous disease the
location of the reference plane will change as the retinal nerve fiber layer will get
progressively thinner. In release 1.11 the standard reference plane is located 50
m posteriorly of the mean contour line height in the segment between -10
degrees and -4 degrees. This segment is located at the papillomacular nerve fiber
bundle, supposed to change in last order during development of glaucoma. In
addition, the range of possible location changes is also minimized using this
segment because the thickness of the RNFL at the papillomacular bundle probably
is only approximately 50 m thick, whereas in the upper and lower temporal
arcuate regions the RNFL is up to 300 µm thick (Quigley & Addicks 1982). This
25
new standard reference plane has been shown to be better in separating subjects
with different stages of glaucoma (Tuulonen et al. 1994)
2.3.2. Evaluation of retinal nerve fiber layer
Using the white light of an ophthalmoscope the healthy nerve fiber bundles
are seen best at the peripapillary retina as silvery striations. RNFL atrophy appears
as a darker red area in which visibility of the normal striation pattern is reduced or
missing. However, a much better visibility of the nerve fiber layer and its defects
can be achieved with a green light. The advantage of red-free ophthalmoscopy was
probably first noted by Vogt in 1917.
A photographic technique is an easier and more permanent way to evaluate
RNFL. Moreover, although many localized sector-shaped retinal nerve fiber layer
defects can be detected with funduscopy, some will be visible only in good
photographs. Diffuse nerve fiber loss is difficult to detect by slit lamp examination
and usually requires photographs. Any defect visible by clinical examination,
however, can generally be demonstrated with photographs.
A monochromatic light was introduced into ophthalmic photography by
Behrendt and Wilson (1965). They used interference filters and black-and-white
film and noticed that the RNFL was invisible for red light, its visibility increased
in green-blue and blue light from 549 to 477 nanometers (nm) and began to
disappear at 431 nm. Delori and Gragoudas (1977) found wavelengths from 475 to
26
520 nm best suited for retinal nerve fiber layer photography. Miller and George
(1978) achieved best results with a 540 nm filter and black-and-white film.
Sommer et al. (1983) could improve the nerve fiber visibility using a 566 nm
short-pass cut-off filter. Airaksinen et al. (1982) and later Peli et al. (1987)
reported easier detection of nerve fiber layer defects with a wide-angle fundus
camera, using high-resolution, fine-grain, black-and-white film with a blue
narrow-band interference filter of 495 nm wavelength (a Wratten # 58). The
computerized image enhancement has been used to improve analysis of the nerve
fiber layer photographs (Peli et al. 1986). In another method the nerve fiber layer
is first photographed on color film using white light, and then the color
transparency is reproduced on black-and-white film through a green filter to
eliminate the disturbing image of deeper retina and choroid (Hoyt et al. 1973,
Frisen 1980). However, Delori et al. (1977) reported that with monochromatic
light, nerve fibers can be observed 2 to 3 times further away from the optic disc
than with white light. In part this is also caused by the higher resolving power of
low-sensitive, black-and-white films in contrast to that of the color films (Ducrey
et al. 1979, Frisen 1980).
Since information given by RNFL photography is qualitative in nature the
grading systems have been developed in attempt to quantify RNFL abnormalities
(Airaksinen et al. 1984, Niessen et al. 1995, Quigley et al. 1993). Quigley and
associates (1993) developed a four level grading system with three features
27
assessed: the brightness of the reflexes, the RNFL texture, and the degree to which
the RNFL obscured the view of retinal blood vessels. Readings were performed by
comparing the areas above and below the optic disc to that directed towards the
fovea. Based on these three features, diffuse atrophy was divided into mild,
moderate and severe grades, and localized (wedge) atrophy was graded in two
levels. For a final code, diffuse and wedge atrophy was combined into a single
code. Milder wedge defects were included with mild diffuse atrophy and more
severe wedge defects with moderate diffuse atrophy.
Niessen et al. (1995) used a visually supported grading system where the
patient’s photograph was compared with set of 25 reference photographs,
numbered from 25 (broad, clearly striated nerve fiber bundles) to 1 (no nerve
fibers visible). Their classification system was based on grading diffuse atrophy,
separately for upper and lower halves.
Airaksinen et al. (1984) used a semiquantitative scoring system where the
optic disc was divided into 10 sectors: two sectors for the papillomacular bundles,
both extending 20 degrees above and below the horizontal, two sectors for the
nasal area, both extending 60 degrees to either side of the horizontal, and three
sectors each for the superior and inferior arcuate areas, each extending 30 degrees
to the nasal side of the vertical meridian. Each sector was scored separately for
localized and diffuse loss of nerve fibers. The localized and diffuse scores were
added together to provide an overall score.
28
2.4. ASSOCIATION OF FUNCTION AND STRUCTURE
In glaucoma, the anatomic loss of neural tissue goes along with deterioration of
function. Quite a number of reports have described a moderate to fairly good
correlation between optic disc, retinal nerve fiber layer and achromatic visual field
parameters (Hart et al. 1978, Balazsi et al. 1984, Airaksinen et al. 1985, Drance et
al. 1986, Guthauser et al. 1987, Jonas et al. 1988, Caprioli & Miller 1988, Funk et
al. 1988 & 1993, Brigatti & Caprioli 1995, Weinreb et al. 1995). Compared with
other disc parameters the disc rim area was found to be most highly correlated
with visual field mean deviation by several investigators (Caprioli & Miller 1988,
Funk et al. 1993, Brigatti & Caprioli 1995). Recently, Tsai et al. (1995)
investigated relationship between both W/W and B/Y visual field sensitivity and
optic disc parameters using a Heidelberg Retina Tomograph (Software version
1.10). They found that peripapillary retinal height, rim area and rim to disc area
were highly correlated also with B/Y visual field mean deviation when both
normal subjects and glaucoma patients were included into the analysis. With only
glaucoma patients in the analysis the peripapillary retinal height was the only
parameter significantly correlated with the visual field mean deviation. Weinreb et
al. (1995) investigated the association between RNFL measurements and W/W
visual field loss in 53 patients with POAG and did not find a significant
association between RNFL cross section area and global measures of visual field
29
loss. Brigatti and Caprioli (1995) evaluated patients with early to moderate
glaucoma using the HRT and W/W perimetry. They found the cup shape measure
to be the only parameter correlating statistically significantly with the visual field
indexes. No significant correlation was found between mean nerve fiber layer
height and the visual field indexes. The discrepancies between the results of these
studies may be explained by the differences in ONH parameters used and
differences in population.
The detectable RNFL and optic disc abnormalities usually precede criteria
referred to as typical glaucomatous field loss in conventional W/W perimetry
(Quigley et al. 1989, Sommer et al. 1991, Quigley et al. 1992, Tuulonen et al.
1993, Zeyen & Caprioli 1993). In early glaucoma the visual field defects develop
often in one hemifield only. Is the “healthy” hemifield really unaffected by the
disease or does the standard perimetry not detect the functional loss associated
with early optic nerve abnormalities? Sommer et al. (1991) found RNFL defects in
both hemispheres of over 80% of eyes when visual field loss was first
documented. In contrast, visual field loss (by kinetic and suprathreshold static
perimetry) involved both hemifields in only 14 % of eyes. This suggests that in
many cases there is a delay until sufficient damage for definitive diagnosis of
glaucoma on the basis of W/W visual fields occurs and the seemingly “healthy”
hemifield may in fact already be affected by glaucoma.
30
3. PURPOSE OF THE PRESENT STUDY
The aim of the present work was to develop a technique for measuring the
absorption of blue light by the lens to correct B/Y perimetry results and to
evaluate associations of optic nerve tomographic measurements as well as semiquantitative RNFL loss scores with visual field results.
The specific topics were:
1. To investigate the relationship between lens transmission indices obtained by
fluorometry and B/Y visual field thresholds in nonglaucomatous individuals (I).
2. To find out if B/Y perimetry results can be corrected better by using lens
fluorometry than age (I).
3. To determine how the B/Y perimetry results adjusted for age and lens correlate
with the optic disc parameters measured with the Heidelberg Retina Tomograph
(II-III).
4. To compare the strength of the association of the optic nerve head
morphological variables with the B/Y and W/W visual field results (II-III).
5. To test the correlation between the semi-quantitative score of RNFL evaluation
and the retinal sensitivity determined with B/Y perimetry (IV).
6. To study the ability of B/Y hemifield tests and the HRT measurements to
separate patients with glaucoma from normal subjects (III).
31
7. To determine how the B/Y visual field, HRT measurements and RNFL scores
label the hemifields found to be normal on W/W perimetry in patients with
glaucoma (III-IV).
Short answers
 page 67
32
4. MATERIAL AND METHODS
4.1. Subjects
One randomly chosen eye of 40 nonglaucomatous subjects with a mean age
of 57 years (range, 29 to 84 years) was evaluated. Criteria for subject eligibility
consisted of normal findings in the ocular examination, a distance refractive error
between + 4 and - 4 diopters spherical equivalent with not more than 1.5 diopters
of cylinder, a normal W/W visual field, no family history of glaucoma, no history
of ocular or neurologic disease, no history of diabetes or other systemic diseases,
and no history of any medications that are known to affect the visual field
sensitivity or color vision. No restriction was placed on visual acuity. The lowest
visual acuity due to reduced transmission properties of the lens was 0.2.
In addition to nonglaucomatous subjects, we examined 37 patients with
mean age 60 years (range, 30 to 82 years). Ten patients had ocular hypertension
(intraocular pressure  22 mmHg on 3 or more occasions) with normal optic disc,
normal retinal nerve fiber layer and normal W/W visual fields. The normalcy of
the optic discs was determined subjectively by experienced observer based on the
shape of the cup, width and contour of the rim and the texture of appearance of the
neural tissue of the optic disc. Twenty three patients had primary open-angle
glaucoma with elevated IOP and glaucomatous optic disc damage and no apparent
contribution from other ocular or systemic disorders. They were divided into three
groups according to W/W visual field defects: 1) early glaucoma in nine patients
33
with a mean deviation (MD) of better than -5 dB, 2) moderate glaucoma in six
patients with MD between -5 and 10 -dB, 3) advanced glaucoma in eight subjects
with MD worse than -10 dB. In addition, four ocular hypertensive patients with
normal W/W visual fields, but abnormal retinal nerve fiber layer and/or optic disc
were added to the early glaucoma group because they can be labeled as patients
with “preperimetric” (Horn et al. 1997) glaucoma. More specific information for
the various patient groups is provided in Table 1.
In publication IV eight nonglaucomatous subjects, 4 patients with ocular
hypertension and 4 patients with glaucoma had to be excluded because of poor or
missing photographs. The remaining 61 patients included 32 normal subjects with
mean age 54 years (range 29 to 83 years) and 29 patients with ocular hypertension
or glaucoma with mean age of 57 years (range 30 to 82 years).
34
Table 1. Characteristics of subjects.
The figures are mean (standard deviation) if not otherwise stated.
Clinical groups:
Normal
N = 40
Age, years
Ocular
Early
Moderate
Advanced
hypertension
glaucoma
glaucoma
glaucoma
N = 10
N = 13
N=6
N=8
57.5 (14.1)
56.3 (17.5)
60.0 (11.7)
61.6 (14.5)
64.9 (13.8)
29-84
30-77
39-78
44-81
43-82
-1.1 (2.6)
-0.8 (1.4)
-2.2 (1.7)
-7.1 (1.8)
-18.4 (9.5)
0 (3.0)
-2.8 (4.0)
-4.2 (4.6)
-7.8 (2.5)
-13.8 (7.3)
Cup shape measure
-.22 (.08)
-.20 (.05)
-.11 (.06)
-.11 (.08)
.01 (.06)
Rim volume, mm3
.39 (.15)
.37 (.14)
.25 (.14)
.28 (.12)
.08 (.06)
Rim area, mm2
1.46 (.34)
1.38 (.33)
1.18 (.42)
1.21 (.37)
.65 (.29)
area, mm2
1.23 (.34)
1.21 (.44)
1.02 (.26)
.93 (.26)
.37 (.27)
Cup/disc area ratio
.21 (.15)
.28 (.17)
.45 (.12)
.33 (.25)
.65 (.14)
Range
MD of W/W
visual field, dB
MD of B/Y
visual field, dB
RNFL cross section
4.2.1. Fluorometry of the crystalline lens (I - IV)
Autofluorescence (AF) of the lens was measured using a fluorometer designed,
built and clinically tested in the Department of Ophthalmology, University of
Oulu. The schematic representation of the optical system of the fluorometer is
presented by Siik et al. (1991). The excitation light source is a 25 W incandescent
lamp from Zeiss biomicroscope fed by a current-regulated power supply. Infrared
light is blocked by an absorption filter. The wavelength of excitation light is set by
an interference filter with peak transmission at 495 nm. A barrier interference
35
filter has peak transmission at 520 nm. Transmission characteristics of these filters
are shown by Siik et al. (1991).
The optical system consist of a moving motorized lens and a fixed lens that
focus the excitation light on the eye and collect emitted light from the eye. The
dimensions of the illumination slit measured at the focal plane in air are 0.1×1.0
mm. The beam is scanned linearly along the optical axis of the eye at an angle of
20° while the subject views a fixation target. Fluorescence emitted from the lens
passes the lens system on the opposite side at a symmetrical angle and is
transmitted through a slit and a barrier filter to a photodiode detector (EG&G
Electro-Optics: UV-040 BG, spectral range 250 nm- 1150 nm). Any blue light due
to scatter within the eye is reflected from the barrier filter and focused in front of
the ocular.
AF measured as a function of the distance in the eye is processed by lownoise amplifiers, digitized, stored and plotted on the computer screen. The device
produces a graphic fluorescence profile which consist of anterior and posterior
juxtacortical peaks and a central plateau (Figure1 in publication I). A transmission
index was calculated from the ratio between the heights of the posterior and
anterior AF peaks. The coefficient of variation (2.9%) for the lens transmission
index was determined previously by Siik et al. (1991).
Each unidirectional scan from the vitreous to the anterior chamber takes
about 3 seconds. No contact lens is used. The instrument is calibrated before each
36
measurement using a fluorescent reference surface. The long-term stability curves
of the entire fluorometer show good stability at two different fluorescence levels
as shown previously by Siik et al. (1991).
4.2.2. Visual field testing (I - IV)
We obtained both W/W and B/Y visual fields on a modified Humphrey Field
Analyzer, model 610, using program 30-2 (Humphrey Instruments, San Leandro,
California, USA). B/Y perimetry was performed with a 100 cd/m2 yellow
background and a size V blue (440 nm) stimulus. In nonglaucomatous subjects the
W/W perimetry was carried out during the first visit.
The calculation of B/Y visual field total and hemifield MD and W/W visual
field hemifield MD was based on results of nonglaucomatous subjects enrolled in
this study. Variability of the static perimeter threshold is known to increase with
distance from fixation (Heijl et al. 1987, Wild et al. 1995). Almost all subjects, and
particularly the elderly subjects felt that B/Y visual field testing was more
distressing. When patient fatigue is a significant factor the variability increases.
The peripheral points are then more easily missed than central ones. Therefore we
decided to use the 24-2 test data to obtain a more precise model of the normal
hemifields. For achieving data for program 24-2 test, we subtracted from the
program 30-2 data all the most peripheral location test values, with the exception
of two nasal points.
37
The age- and lens-adjusted mean deviation values of B/Y visual field for
total visual field (II, IV) and separately for superior and inferior hemifields (III,
IV) were calculated from the B/Y visual field (program 24-2) total and hemifields
mean sensitivity (MS) values as follows. First, a linear regression model was fitted
in the normal subjects, in which the dependent variable was the B/Y visual field
respective MS and the regressor variables were age and LTI. The expected mean
sensitivity (EMS) was obtained from the estimated regression coefficients a, b,
and c as EMS = a + b × LTI + c × age. Second, the age- and LTI-adjusted mean
deviation was calculated as the difference between the actually obtained and
expected MS values (MD = MS - EMS) for each subject in all clinical groups.
Similarly, the age- and lens-adjusted MD values for superior and inferior
hemifields of W/W visual field were calculated (III, IV). Mean deviation of W/W
total visual field was obtained using the statistical package provided by Humphrey
Instruments (I-IV).
Patients with ocular hypertension and glaucoma were generally more
experienced with W/W visual field testing, but all nonglaucomatous subjects had
had at least one previous experience with automated W/W visual field testing. In
nonglaucomatous subjects the W/W perimetry was carried out during the first
visit. No one had previous experience with B/Y visual field testing.
4.2.3. Measurements of the optic disc (II - III)
38
The Heidelberg Retina Tomograph (Heidelberg Engineering GmbH, Heidelberg,
Germany) with software version 1.11 was used to acquire and evaluate
topographic measurements of the optic disc of all the subjects. The Heidelberg
Retina Tomograph (HRT) is a confocal imaging device that uses a diode laser at
670 nm as a light source. The system consists of the laser scanning camera,
mounted on a standard ophthalmic stand with chin rest, the operation panel, and a
personal computer system.
A three-dimensional image was acquired as a series of optical section
images at 32 consecutive focal planes. Image acquisition time was approximately
1.6 seconds. The instrument does not require dilatation of the pupil (Lusky et al.
1993). The three 10 images were obtained for each eye, and the mean image of
the three scans was used for the optic disc structure measurements. Optic disc
contour line was manually marked around the disc on the HRT screen with the
mouse: the inner edge of the scleral ring (Elschnig´s ring) corresponding the inner
edge of the contour line. The determination of the reference plane for each
individual eye was based on the height of the retinal surface at the pappilomacular
bundle. The standard reference plane is located 50 m posteriorly of the mean
contour line height in the segment between -10 degrees and -4 degrees. In
publication II the following information was collected: disc area (DA), cup area
(CA), rim area (RA), cup volume (CV), rim volume (RV), mean depth below
curved surface (MDC), maximum cup depth (MxCD), cup shape measure or third
39
moment (CSM), height variation along contour line (HVC), cup/disc area ratio
(CDR), mean retinal nerve fiber layer thickness (mRNFLt), and retinal nerve fiber
layer cross section area (RNFLcsa). In publication III the superior and inferior
segments were defined in the configuration file with the angles 0 - 180 and 180360, respectively, and the following information was collected: disc area, cup area,
rim area, cup volume, rim volume, cup shape measure, mean RNFL thickness, and
RNFL cross section area. All parameters were corrected for refractive error and
actual radius of corneal curvature (Zinser et al. 1989).
The DA was defined as the total area within the contour line. The CA was
defined as the total area of those parts within the contour line that are located
below the reference plane. The RA was defined as the difference between DA and
CA. The CV was defined as the total volume of those parts within the contour line
that are located below the reference plane. The RV was defined as the total volume
of those parts within the contour that are located above the reference plane (the
reference plane was used as the lower limit for the measurement). The MDC was
defined as mean cup depth below curved surface. The MxCD was defined as the
mean depth of the 5% of pixels with the highest depth values within the contour
line (the depth was determined relative to the curved surface). The CSM was
defined as the frequency distribution of depth values relative to the curved surface
of those parts located inside the contour line. The HVC was defined as the
difference in height between the most elevated and the most depressed point of the
40
corrected contour line. The mRNFLt was defined as reference height minus the
mean height of contour. The RNFLcsa was defined as the mRNFLt times the
length of the contour line.
The HRTCALC Utility Version 1.03 (Heidelberg Engineering GmbH,
Heidelberg, Germany) was used to calculate the parameter values for the HRT
mean images. The data was transported as an ASCII data file to the SPSS 6.1.2.
for Windows (SPSS Inc. Chicago, Illinois).
4.2.4. Evaluation of the retinal nerve fiber layer (IV)
RNFL photographs were taken with a 60 wide-angle fundus camera (Canon, Inc,
Kawasaki City, Japan) with a monochromatic blue interference filter on low
sensitivity black-and-white film. The technique of RNFL photography has been
reported earlier in detail by Airaksinen & Nieminen (1985). The RNFL loss was
assessed using the semi-quantitative scoring method by experienced observer in a
completely masked fashion with the optic discs covered. The photographs were
analyzed in a random order without any clinical information. It has been shown
(Airaksinen et al. 1984) that the reproducibility of the method is good (CV=0.15).
To localize the nerve fiber damage at the optic disc margin, the
circumference of the optic nerve head was divided into ten sectors (Figure 1 in
publication IV). Each sector was scored from 0 to 3 separately for localized and
diffuse loss of nerve fibers (with 0 = no damage; 1 = mild atrophy; 2 = moderate
41
atrophy; and 3 = severe atrophy). Diffuse RNFL loss was evaluated by the
segments of inferior arcuate area, superior arcuate area, papillomacular area, the
nasal area. When a segment was found to have diffuse loss the extent of loss in
each sector was determined and the respective score given to each sector of the
segment. From these data, a total localized score, a total diffuse score and a total
overall score were calculated. The total localized score is the sum of the localized
scores in each of the ten sectors. The total diffuse score is the sum of diffuse
scores in each of the ten sectors. The total overall score is the sum of the total
localized score and the total diffuse score. The RNFL loss overall scores were
calculated separately for the superior (segments 1 - 4 and 10) and the inferior
(segments 5 - 9) hemispheres.
4.2.5. Data analysis
The relation among lens transmission index, B/Y visual field threshold values and
age was analyzed using regression technique. The variability of the B/Y mean
sensitivity by age and lens transmission index in healthy subjects was evaluated
using a multiple regression analysis with MS as the dependent variable and age
and lens transmission index as the independent variables (I).
In publication II the associations of the optic nerve head (ONH)
morphological measures with the mean deviation of B/Y and W/W visual field
were evaluated by quadratic regression models in which each of the ONH
42
structural variables was in turn the outcome variable and the B/Y visual field MD
(adjusted for age and LTI) and the W/W visual field MD were the regressor
variables. Both the linear and quadratic term of the regressors were included in the
models to allow for significant departures from linearity in the association. The
multiple correlation coefficient from a model including either B/Y or W/W visual
field MD (linear and quadratic term) were used to describe the strength of the
association between the ONH morphological variable and the visual field defect
measure in question. Forward stepwise procedure was used to find the order of
significance of the B/Y and W/W visual field mean deviation variables.
The linear correlation coefficients (Pearson’s R) were calculated between
the functional and structural variables to evaluate the strength of the following
associations: 1) hemifields of B/Y and W/W visual field with the respective HRT
parameters (III); 2) B/Y and W/W visual fields with the RNFL (IV).
In order to evaluate the strength of the association of B/Y and W/W visual
fields, respectively, with the HRT parameters (II, III) and RNFL findings (IV) in
the early stages of glaucoma, a further analysis was made in which the patient
group with advanced glaucoma was excluded and linear correlation coefficients
between the visual field measures and morphological variables were calculated.
The W/W hemifield was defined to be “normal” when age- and lensadjusted hemifield MD was better than - 2 dB. In publication IV an additional
analysis was made with B/Y visual field and RNFL evaluation data corresponding
43
to “normal” W/W hemifield (N=19) in patients with early glaucoma. Their
hemifield MD values of B/Y and W/W visual fields were correlated (Pearson’s R)
with the corresponding hemisphere overall scores of RNFL loss.
In publication III a logistic regression by forward stepwise method was
performed to investigate the sensitivity and specificity in determining the
glaucomatous fields as abnormal and visual fields of healthy individuals as
normal. Sixty three W/W hemifields of normal subjects with normal HMD and 33
W/W hemifields of glaucoma patients with abnormal HMD (worse than -2 dB)
were analyzed with respective B/Y and HRT data. A further analysis was made to
find out whether the normal hemifields of glaucoma patients examined with W/W
perimetry are labeled normal also by B/Y and HRT measurements, and whether
abnormal W/W visual fields are classified abnormal also by B/Y and HRT
measurements. For these purposes forward stepwise analysis was made where
normal W/W hemifields (N= 21) of glaucomatous subjects were considered
normal and abnormal W/W hemifields (N=33) of glaucomatous subjects abnormal.
To determine whether there is a statistically significant difference in the
hemifield MD of B/Y visual fields, obtained from “normal” W/W hemifields, in
normal subjects, patients with ocular hypertension and patients with early
glaucoma an analysis of variance with Duncan’s multiple range test was
performed (IV).
44
Statistical analysis was performed using program SPSS 6.1.2. for Windows
(SPSS Inc. Chicago, Illinois).
45
5. RESULTS
5.1. Correction of perimetry results
The total MS values of the visual field programs 30-2 and 24-2 as well as the MS
of 9 eccentricity pattern showed a statistically highly significant correlation with
the lens transmission index in non-glaucomatous individuals (R = 0.81, 0.83 and
0.86, respectively; P < 0.0001). The correlation of B/Y visual field (program 242) MS values in nonglaucomatous individuals plotted against the lens transmission
index is shown in figure 1. It is a linear relationship with the data best fitted by
the equation: Y= 8.56 + 0.173 × X where X is the lens transmission index and Y
the expected total MS in healthy individuals.
Age showed a good correlation to the B/Y MS values (program 24-2) and
the LTI (R = - 0.77, = 0.80, respectively; P < 0.0001) in non-glaucomatous
subjects. However, the regression analysis with MS value as the dependent
variable indicated that lens transmission index provided a more precise prediction
of the MS value than age. The residual standard deviation (SD) of the regression
model including age alone was 0.44 dB larger than that having lens transmission
index as the independent variable ( SD 3.66 dB and 3.22 dB, respectively). Adding
the lens transmission index to the model including only age reduced the residual
standard deviation by 0.57 dB to 3.09 dB (P = 0.003), but the residual standard
deviation was reduced only by 0.13 dB (P = 0.04) when age was added to the
model containing lens transmission index only.
46
Although the reference level for correcting B/Y perimetry results can be
determined more precisely using fluorometry of the lens than age alone, the
variability was further decreased when both variables, LTI and age were used.
Therefore, in consequent publications the age- and LTI-adjusted MD of total
visual field and hemifields was calculated as the difference between the expected
and measured MS values. The fitted regression line for the expected B/Y visual
field (program 24-2) total mean sensitivity values (EMS) in normal subjects was:
EMS = 19.31 + 0.123 × LTI - 0.122 × age (years). The standard errors of these
coefficients were 5.32, 0.030, and 0.058, respectively. The residual standard
deviation about the line was 3.09. The r-square value of the model was 0.72. The
fitted regression line for the expected B/Y superior hemifield mean sensitivity
values (EHMS) in normal subjects was: EHMS = 19.38 + 0.114 × LTI - 0.131 ×
age (years), and for the inferior hemifield: EHMS = 19.23 + 0.133 × LTI - 0.113 ×
age. In publications III and IV the age- and LTI-adjusted hemifield MD was
calculated also for W/W visual field. The fitted regression line for the expected
W/W superior HMS values in normal subjects was: EHMS =25.65 + 0.072 × LTI 0.058 × age, and for the inferior hemifield: EMS = 28.12 + 0.063 × LTI - 0.08 ×
age.
When all subjects (N=77) were included into the analysis there was a strong
correlation between the age- and LTI-adjusted B/Y visual field and W/W visual
field total MD values (R = 0.78) (Fig. 2) as well as between the age- and LTI-
47
adjusted B/Y and W/W visual field superior HMD values (Pearson´s R = 0.79) and
inferior (Figure 1 in publication III) HMD values (R = 0.81), respectively.
5.2. Association with scanning confocal laser optic disc measurements
5.2.1. Mean deviation of B/Y and W/W total visual field (II)
The W/W visual field mean deviation values and the age- and LTI-adjusted MD
values of the B/Y visual fields both showed the highest correlation with the cup
shape measure (the multiple correlation coefficients from quadratic regressions
were 0.65 and 0.65, respectively) (Figs 3 and 4).
The structural variables under study can be divided into three broad
categories with respect to their observed relations to the visual field measures
(Table 2). In the first class (1) were the cup shape measure and the mean cup depth
below curved surface which were more strongly associated with the B/Y visual
field MD than with the W/W visual field MD. However, in these cases, also W/W
terms were selected by the stepwise algorithm. Of all structural parameters
considered, the linear term of the B/Y visual field MD had the strongest
correlation with the cup shape measure (linear R = 0.65). Upon that both the linear
and quadratic term of the W/W visual field MD were significantly associated with
the cup shape measure, but the quantitative increment in the value of the multiple
correlation coefficient due to W/W perimetry results was small (multiple R = 0.70
after including both W/W terms).
48
The second class (2) consisted of variables which were also more correlated,
but not substantially so, with the B/Y visual field than with the W/W visual field,
such that the linear B/Y term was the only functional measure selected by the
stepwise procedure. These parameters were cup to disc area ratio, rim area, rim
volume, cup area, cup volume and maximum cup depth.
In the third (3) category we had three structural variables which were more
correlated with the W/W visual field MD than with the B/Y visual field MD. The
linear term of the W/W visual field MD had the statistically strongest correlation
with the retinal nerve fiber layer cross section area (linear R = 0.54) (Figure 4 in
publication II), mean retinal nerve fiber layer thickness (linear R = 0.53) and
height variation along the contour line (linear R = 0.48). The quadratic term of the
W/W visual field MD added some contribution with the RNFLcsa and mRNFLt
upon the linear term (multiple R = 0.61 and R = 0.62, respectively). In all these
three variables addition of the B/Y terms did not improve the fit of the regression.
49
Table 2. Multiple correlation coefficients of B/Y and W/W Visual Field Mean
Deviations (MD), respectively, with Tomographic Data from quadratic regression
models (+ or - sign in parentheses indicate the direction of the linear component of
the association). (N = 77)
Tomographic Parameter
1
MD of
MD of
B/Y visual field
W/W visual field
Cup shape measure
(-).65
(-).65
Mean cup depth below
(-).35
(-).16
Cup/disc area ratio
(-).59
(-).56
Rim area, mm2
(+).54
(+).49
Rim volume, mm3
(+).54
(+).52
Cup area, mm2
(-).52
(-).50
Cup volume, mm3
(-).34
(-).29
Maximum cup depth, mm
(-).30
(-).25
RNFL cross sectional area,
(+).48
(+).61
Mean RNFL thickness, mm
(+).49
(+).62
Height variation along
(+).44
(+).48
curved surface, mm
2
3
mm2
contour line, mm
Scatterplots (Figs. 2 - 4) show that the advanced glaucoma group greatly
influenced the associations. Moreover, it was this patient group that made the
associations of W/W visual field values with the HRT parameters to deviate from
a linear pattern. When subjects with advanced glaucoma were excluded from the
analysis the associations were linear, and the correlation coefficients (ordinary
Pearson’s R) decreased considerably but more so for the W/W perimetry results
(Table 3).
50
Table 3. Correlation coefficients (Pearson’s R) of B/Y and W/W Visual Field Mean Deviations
(MD), respectively, with Tomographic Data (the 8 subjects with advanced glaucoma excluded, N
= 69).
Tomographic Parameter
MD of
B/Y visual field
-.46
MD of
W/W visual field
-.37
1
Cup shape measure
-.28
-.12
2
Mean cup depth below curved
surface, mm
Cup/disc area ratio
-.45
-.30
Rim area, mm2
.37
.17
Rim volume, mm3
.39
.27
Cup area, mm2
-.43
-.33
Cup volume, mm3
-.28
-.20
Maximum cup depth, mm
-.14
-.02
RNFL cross section area, mm2
.27
.37
Mean RNFL thickness, mm
.31
.40
Height variation along contour
line, mm
.14
.19
3
5.2.2 Mean deviation of hemifield (III)
Among HRT parameters the cup shape measure (CSM) showed the highest
correlation with the respective HMD values of all W/W and B/Y visual field
hemifields. The highest correlation (Pearson’s R = - 0.62) was found between B/Y
visual field inferior HMD and CSM obtained from the superior part of the optic
disc (Fig. 5). Rim volume, rim area, cup area, mean RNFL thickness and RNFL
cross section area also showed a statistically significant correlation with respective
HMD values of the W/W and B/Y visual field. The results of analysis with all
subjects (N=77) included are summarized in Tables 4 and 5. Without the advanced
glaucoma subjects the correlation coefficients decreased considerably but more so
for the W/W perimetry data (Tables 3 and 4 in publication III).
51
Table 4. Correlation coefficients (Pearson’s R) of B/Y and W/W visual field superior hemifield
mean deviations (HMD), respectively, with tomographic data of the inferior part of the optic disc
in normal controls (n=40), patients with ocular hypertension (n=10) and glaucoma (n=27).
Tomographic Parameter
of the inferior part of
optic disc
Disc area, mm2
HMD of B/Y
visual field
superior hemifield
-.03§
HMD of W/W
visual field
superior hemifield
-.00§
Cup area, mm2
-.52 
-.48 
Rim area, mm2
.53 
.51 
Rim volume, mm3
.56 
.50 
Cup shape measure
-.56 
-.57 
Mean RNFL thickness, mm
.47 
.53 
RNFL cross section area, mm2
.47 
.54 
 p<0.0001; § p> N.S. (not significant)
Table 5. Correlation coefficients (Pearson’s R) of B/Y and W/W visual field inferior
hemifield mean deviations (HMD), respectively, with tomographic data of the superior part of
the optic disc in normal controls (n=40), patients with ocular hypertension (n=10) and
glaucoma (n=27).
Tomographic Parameter
of the superior part of
optic disc
Disc area, mm2
HMD of B/Y
visual field
inferior hemifield
-.01§
HMD of W/W
visual field
inferior hemifield
-.03§
Cup area, mm2
-.45 
-.39 
Rim area, mm2
.49 
.40 
Rim volume, mm3
.42 
.40 
Cup shape measure
-.62 
-.56 
Mean RNFL thickness, mm
.39 
.47 
RNFL cross section area, mm2
.38 †
.43 
 p<0.0001; † p<0.01; § p> N.S.
52
5.2.3. Sensitivity and specificity (III)
The logistic regression analyses with B/Y HMD values and HRT parameters as
covariates were performed to test the sensitivity and specificity of the model to
correctly identify normal W/W hemifields of the healthy individuals and
glaucomatous W/W hemifields of the glaucoma patients. The B/Y term was
selected by the first step showing a sensitivity of 93.7% , specificity of 84.9% and
overall correct identification of 90.6%. Four of 63 hemifields of normal subjects
were considered as abnormal and 5 of 33 abnormal W/W hemifields were
considered on bases of B/Y results as normal. The cup shape measure (CSM) term
was entered on the second step. The sensitivity was 95.2% with B/Y and CSM
term together in the model, specificity was 87.9 %, and the overall correct
identification was 92.7%. Other HRT terms did not add significantly to the model.
5.2.4. Normal hemifield by W/W perimetry (III)
In glaucoma subjects (N=27) the superior W/W hemifield was normal (HMD
better than - 2 dB) in 7 cases and the inferior hemifield in 14 cases. The superior
W/W hemifield was abnormal (HMD worse than -2 dB) in 20 cases and the
inferior hemifield in 13 cases. The B/Y term was entered on step one when the
glaucoma patient´s normal (N=21) and abnormal hemifields (N=33) were analyzed
by forward stepwise method with the B/Y hemifield mean deviation values and the
HRT parameters as covariates. The B/Y hemifield mean deviation data labeled as
53
abnormal 8 of 21 (38%) cases with normal W/W hemifield. None of the HRT
parameters added significantly to the model.
With only the HRT parameters in the model the CSM was entered on step
one. This variable classified 11 of 21 (52%) patients with normal W/W hemifields
as abnormal. Other HRT parameters did not add significantly to the model. Five of
33 with abnormal W/W hemifields were considered normal by the B/Y hemifield
mean deviation as well as the CSM.
5.3. Association with semi-quantitative RNFL loss scores
5.3.1. Mean deviation of B/Y and W/W visual field (IV)
The MD values of B/Y and W/W total visual field (program 24-2) were both
statistically highly significantly related to the total diffuse nerve fiber layer loss
score (Pearson´s R = - 0.73 and - 0.71, respectively; P < 0.0001) when all subjects
(N=61) were included into the analysis. The correlation of B/Y visual field MD
values plotted against the total diffuse nerve fiber loss score is shown in figure 6.
A statistically significant correlation was also found between the B/Y and W/W
visual field MD values and the total overall nerve fiber layer loss score (Pearson´s
R = - 0.71 and - 0.66, respectively; P < 0.0001). No correlation was found between
MD values and total localized score of the RNFL loss.
When eight subjects with advanced glaucoma were excluded from the
analysis (N= 61 - 8 = 53) the correlation coefficients decreased but more so for the
54
W/W perimetry results. The Pearson’s correlation coefficients between MD values
of B/Y and W/W total visual field and total diffuse score of nerve fiber loss were 0.60 (P < 0.0001) and - 0.50 (P = 0.001), respectively.
5.3.2. Normal hemifield by W/W perimetry (IV)
The age- and lens-adjusted mean deviation of W/W hemifield was better than - 2
dB in 55 hemifields of normal subjects (N=32), in 12 hemifields of patients with
ocular hypertension (N=6) and in 19 hemifields of patients with early glaucoma
(N=12). None of the patients in the moderate and advanced glaucoma groups had a
W/W hemifield better than - 2 dB. The age- and lens-adjusted MD of B/Y
hemifield was worse than - 5 dB in 5 of 55 (9%) hemifields of normal subjects
considered to be “normal” on W/W perimetry, in 4 of 12 (33%) ocular
hypertensive patients, and in 10 of 19 (53%) early glaucoma patients (Fig. 7). In
the 4 hemifields (B/Y HMD worse than - 5 dB) of the ocular hypertensive patients
the RNFL score showed no abnormality (Fig. 8).
In patients with early glaucoma the RNFL overall loss score indicated
abnormality in 16 of 19 (84%) hemispheres which corresponded to “normal” W/W
hemifield (Fig. 8). The RNFL score was abnormal also in five hemispheres
corresponding to respective B/Y hemifields with an MD better than zero in
patients with early glaucoma. The overall score of RNFL loss in the hemispheres
(N=19) was well correlated with the HMD value of the B/Y mirror-hemifields,
55
found to be “normal” on W/W perimetry (Pearson’s R = - 0.56, P = 0.012). The
respective Pearson’s correlation coefficient for W/W hemifield mean deviation
was - 0.26 (P = 0.28).
The analysis of variance with Duncan’s multiple range test showed a
statistically significant difference (P = 0.0002) between the HMD values of B/Y
perimetry obtained from “normal” W/W hemifields of normal subjects (N = 55)
(0.39 dB ± 2.95) and ocular hypertensive patients (N = 12) (-2.52 dB ± 4.34) (Fig.
7). The respective B/Y hemifield data of early glaucoma patients (N = 19) (-3.56
dB ± 4.99) was not statistically significantly different from that of the patients
with ocular hypertension.
56
6. DISCUSSION
6.1. Correction of B/Y perimetry results for lens yellowing and age (I)
The major disadvantage of B/Y perimetry is that transmission properties of the
lens should be measured in order to separate optical and neural sources of short
wavelength sensitivity loss. This study presents a procedure that provides a
practical estimate of absorption of the blue light in an individual lens.
The scatterplot of the ocular hypertensive patients (Fig. 9) shows that in 11
of 15 cases the MS values fall inside the limits of the 95% prediction interval
suggesting that their B/Y mean sensitivity values are not different from that of the
normals. Without the knowledge of the lens transmission properties many of the
lower MS values could erroneously have been interpreted as abnormal. Given that
the ocular hypertensive patients have normal B/Y visual fields they would have
only a 5% change of being outside of the limits of the 95% prediction interval.
The MS values in the early glaucoma group were divided between the ones
inside (5 patients) and the ones outside (4 patients) of the prediction interval of the
normals (Fig. 10). This suggests that on the basis of the MS values of their B/Y
visual fields the 5 patients within the prediction interval are not different from our
normals or most of our ocular hypertensives patients. However, the 4 patients with
MS values below the lower reference level would be glaucomatous measured both
with W/W perimetry and with B/Y perimetry with lens autofluorescence taken into
account. In the moderate and advanced glaucoma groups 14 of 15 patients had an
57
MS value below the lower reference limit suggesting that they indeed have a true
optic nerve damage and their low B/Y perimetry thresholds are not determined by
lens yellowing alone.
Previous reports have shown that there is a good correlation between lens
transmission values and age (Zeimer & Noth 1984, Siik et al 1991). From this
point the question arose whether performing the lens transmission property
measurement adds information beyond that provided by age alone. To examine
this the regression model was fitted to the perimetry data with both age and lens
transmission index together and separately. The residual standard deviation of the
B/Y mean sensitivity with the lens transmission index alone was smaller than with
age alone. Adding the lens transmission index to the regression model reduced the
residual standard deviation considerably more than adding age to the model.
Therefore the B/Y mean sensitivity data is less variable with the lens transmission
index and we can conclude that the lens corrected MD is more precise than the age
corrected MD.
A good correlation between lens transmission values and age was found
also in this study. However, the interindividual variation of the lens transmission
properties increases considerably with age (Fig. 11). Therefore, the average values
will not serve well for individual cases. To illustrate this statement the MD for
patients with glaucoma using two different equations was calculated. If the two
techniques were identical the MD values in figure 12 would be on or close to the
58
diagonal. Figure 12 shows, however, that thresholds in cases in whom the
individual lens transmission properties were better than expected according to age
the age corrected mean deviations tend to be underestimated. And reverse, if the
patients lens transmission properties are worse than expected the age corrected
MD values tend to be overestimated.
One might postulate that before being able to decide whether a retinal
sensitivity to blue stimulus is abnormal, not only the lens transmission properties
but also the physiologic behavior of retina with age must be taken into account.
According to previous studies the age related decrease in sensitivity of shortwavelength-sensitive cone pathways is 0.05 - 0.15 log units per decade (Johnson
et al. 1988). However, regarding the extent of obtained prediction intervals (i.e.
interindividual variation) the possible influence of neural aging to threshold
estimation seems to be quite small. Actually, the correlation between the lens
transmission index and the MS includes indirectly the aging factor as the older
subjects have lower lens transmission values.
The regression of B/Y MS value on lens transmission index in the normals
can be used to estimate the LTI adjusted MD of B/Y perimetry in patients with
ocular hypertension and glaucoma. However, it was evident that the variability
was further decreased when both variables, LTI and age were used. Therefore, in
consequent publications (II-IV) the age- and LTI-adjusted MD values were
calculated.
59
60
6.2. Association of B/Y and W/W visual field with optic disc and RNFL
In glaucoma, the anatomic loss of neural tissue goes along with deterioration of
function. The relationship between optic disc, retinal nerve fiber layer and
achromatic visual field parameters has been described by several investigators
(Hart et al. 1978, Balazsi et al. 1984, Airaksinen et al. 1985, Drance et al. 1986,
Guthauser et al. 1987, Jonas et al. 1988, Caprioli & Miller 1988, Funk et al. 1993,
Brigatti & Caprioli 1995, Weinreb et al. 1995). Compared with other disc
parameters the disc rim area was found to be most highly correlated with visual
field mean deviation (Caprioli & Miller 1988, Funk et al. 1993, Brigatti &
Caprioli 1995). The correlation coefficient between disc rim area and W/W visual
field mean deviation determined in this study was in concordance with previously
reported results.
Several studies have described a correlation between quantitative
measurements of the optic disc and white-on-white (Brigatti & Caprioli 1995,
Weinreb et al. 1995, Iester et al. 1997) as well as blue-on-yellow (Tsai et al. 1995).
Tsai et al. reported that rim area, rim to disc area and peripapillary retinal height
measured using a Heidelberg Retina Tomograph (Software version 1.10) were
highly correlated with B/Y visual field mean deviation when both normal subjects
and glaucoma patients were included into the analysis. Spearman rank-order
correlation coefficients for B/Y visual field MD were 0.56, 0.45, and 0.51,
respectively. The multiple correlation coefficients determined in our study
61
between B/Y visual field MD and rim area, rim to disc area and mean RNFL
thickness were 0.54, 0.59 and 0.49, respectively. Interestingly, in both studies the
RNFL parameters correlated slightly better with the W/W perimetry results while
the optic disc parameters (rim area and rim to disc area ratio) were better
correlated with B/Y perimetry results.
In this study the B/Y perimetry results correlated well with the optic nerve
head parameters measured with a confocal scanning laser tomograph (II-III) and
with semi-quantitative score of RNFL loss (IV). In view of some previous studies
(Hart et al. 1990, Sample & Weinreb 1992, Sample et al. 1993, Johnson et al.
1993), indicating that B/Y perimetry may precede W/W visual field loss, we
expected the B/Y visual field MD value to be correlated better with the structural
measures than the W/W visual field index. On one hand, the results of this study
indicate that with all five clinical groups included into the analysis the superiority
of the B/Y perimetry over W/W perimetry is not distinct. On the other hand, it was
obvious from the scatterplots that the advanced glaucoma group greatly influenced
the associations. Without the advanced glaucoma subjects the differences of the
correlation coefficients of B/Y and W/W visual fields with the structural
parameters increased (II-IV) suggesting that B/Y perimetry might add information
beyond that of W/W perimetry particularly in early stages of glaucoma.
The statistically significant correlations between structural and functional
characteristics were obtained in this study. However, statistically significant
62
correlations do not necessarily mean that it is possible to accurately predict one
value on the basis of another. The highest coefficient of correlation (R= 0.65)
between ONH and B/Y visual field global parameters (II) yields an R 2 of 0.423,
meaning that the one variable can only account for less than 43% of the variance
observed for the other variable. When the advanced glaucoma patients were
eliminated from the data set, correlations were considerably reduced. The highest
correlation was R= 0.46, which yields an R2 of approximately 0.21. This means
that nearly 80% of the variance of one variable cannot be accounted for on the
basis of the other variable. Thus, in early stages of glaucoma, the ability to predict
functional properties on the basis of structural measures or vice versa, is quite
poor. There is a large interindividual variation among the normal subjects and
patients with less severe glaucoma, and the assessment of only one parameter,
describing either function or structure, could lead to wrong interpretations. For
providing a full characterization of glaucomatous damage the structural and
functional measures are both important. In this study, the B/Y hemifield mean
deviation combined with the CSM of the Heidelberg Retina Tomograph (III)
showed a good sensitivity (95.2%) and specificity (87.9%) correctly identifying
92.7% of the normal and abnormal hemifields classified by W/W perimetry.
6.3. Normal W/W visual field and glaucoma
63
The detectable RNFL and optic disc abnormalities usually precede criteria
referred to as typical glaucomatous field loss in conventional W/W perimetry
(Airaksinen & Heijl 1983, Sommer et al 1991, Tuulonen et al. 1993) On the other
hand, it has also been found (Sample & Weinreb 1992, Johnson et al. 1993) that
B/Y visual field defects occur earlier and are larger than the W/W visual field
defects, suggesting that there is a delay until sufficient damage for diagnosis of
glaucoma on the basis of W/W visual fields occurs.
The scatterplots (Figs. 2, 4) show that 8 of 13 ( 62%) patients with early
glaucoma, 3 of 10 ( 30%) patients with ocular hypertension but only 4 of 40 (10%)
normal subjects had an age- and LTI-adjusted MD of B/Y visual field of less than
-5 dB. While in scatterplots with the W/W visual field MD the ocular hypertension
and early glaucoma patients by definition were clearly separated from the patients
with moderate glaucoma, in the scatterplots with B/Y MD this separation was no
longer distinct. In fact, the B/Y MD of some patients with ocular hypertension and
early glaucoma approached that of the patients with moderate glaucoma.
In early glaucoma the visual field defects develop often in one hemifield
only. Is the “healthy” hemifield really unaffected by the disease or does
conventional perimetry not detect the functional loss associated with early optic
nerve abnormalities? In this study, among the glaucoma patients there were 21
apparently normal hemifields examined with W/W perimetry. With forward
stepwise logistic regression analysis (III) using B/Y hemifield data 38% of these
64
normal hemifields were classified abnormal. With the CSM alone in the model
52% of the cases were classified abnormal.
An abnormal nerve fiber layer (IV) was found in 84 % of hemispheres
respective to the “normal” W/W hemifields in early glaucoma patients. Further,
there was a good correlation (Pearson’s R = - 0.56) between their nerve fiber loss
score and mirror-image hemifield MD values of B/Y perimetry. This indicates that
in the progression of glaucoma the changes of B/Y visual field and the RNFL
follow more closely each other than do changes of the W/W visual field. Although
the chosen criterion for W/W hemifield normality or abnormality is conservative
results of this study suggest that in many cases the seemingly “healthy” W/W
hemifield may in fact already be affected by glaucoma.
6.4. Patients with ocular hypertension and B/Y perimetry
Ocular hypertensive subjects with normal optic disc, normal RNFL and normal
W/W visual fields are considered nonglaucomatous. The analysis of variance (IV)
showed a statistically significant difference between the hemifield MD values of
B/Y visual field in normal and ocular hypertensive subjects, but no difference was
found between ocular hypertensive patients and the early glaucoma group.
Experimental studies have shown that localized RNFL defects can be clinically
detected if more than 50 % of the thickness of the retinal nerve fiber layer is lost
(Quigley & Addicks 1982). Therefore, the patients with ocular hypertension may
65
have an impaired function of the ganglion cells in spite of clinically normal
looking RNFL. However, the results of this study also suggest that while the B/Y
perimetry may in some cases reveal functional damage even before RNFL loss
becomes clinically detectable there are other patients in whom the glaucomatous
RNFL abnormalities can be detected in spite of normal blue-on yellow and whiteon-white visual fields. Long-term follow-up will show whether functional and
structural abnormalities detectable also by other methods will develop in our
ocular hypertensive patients, currently labeled as nonglaucomatous.
66
7. SUMMARY AND CONCLUSIONS
Purpose of this study was to test the applicability of lens autofluorescence
measurements in correcting blue-on-yellow perimetry results for lens yellowing,
and to evaluate the relationship between quantitative ONH, semi-quantitative
RNFL and B/Y visual field test results in normals and patients with glaucoma and
ocular hypertension. One randomly chosen eye of 40 normal subjects and 37
patients with ocular hypertension and different stages of glaucoma was evaluated.
B/Y mean sensitivity (MS) was statistically highly significantly correlated
with lens transmission index (LTI) expressed as measure of the lens
autofluorescence in normal subjects. The regression analysis with MS value as the
dependent variable indicated that lens transmission index provided a more precise
prediction of the MS value than age. The residual standard deviation of the
regression model including age alone was 0.44 dB larger than that having lens
transmission index as the independent variable (3.66 dB and 3.22 dB,
respectively). However, the variability was further decreased when both variables,
LTI and age were used. Therefore, in evaluating the associations of B/Y visual
field with optic disc and retinal nerve fiber layer (RNFL) the age- and LTIadjusted mean deviation (MD) of total visual field and hemifields was calculated
as the difference between the expected and measured MS values.
The B/Y visual field total and hemifield MDs showed a statistically
significant correlation with respective optic nerve head (ONH) parameters such as
67
cup shape measure (CSM), rim volume, rim area, mean RNFL thickness and
RNFL cross sectional area as well with RNFL loss scores. The superiority of the
B/Y perimetry over W/W perimetry was not distinct when all clinical groups were
included into the analysis. Without the advanced glaucoma subjects the
differences of the correlation coefficients of B/Y and W/W visual fields with the
structural parameters increased suggesting that B/Y perimetry might add
information beyond that of W/W perimetry particularly in early stages of
glaucoma.
On the other hand, the ability to predict functional properties on the basis of
structural measures or vice versa, particularly in early stages of glaucoma, is quite
poor, suggesting that for providing a full characterization of glaucomatous damage
the structural and functional measures are both important. In this study, the B/Y
hemifield mean deviation combined with the cup shape measure of the Heidelberg
Retina Tomograph showed a good sensitivity (95.2%) and specificity (87.9%)
correctly identifying 92.7% of the normal and abnormal hemifields classified by
W/W perimetry.
In many cases the detectable changes of B/Y visual field, ONH and RNFL
seem to precede changes of the W/W visual field. With forward stepwise logistic
regression analysis using B/Y hemifield data 38% of the glaucoma patient´s
“normal” W/W hemifields were classified abnormal. With the CSM alone in the
model 52% of the cases were classified abnormal. Moreover, the B/Y hemifield
68
data obtained from “normal” W/W hemifields of early glaucoma patients were
well correlated with respective RNFL loss scores found to be abnormal in 84% of
hemispheres.
The analysis of variance showed a statistically significant difference
between the hemifield MD values of B/Y perimetry obtained from “normal” W/W
hemifields of normal subjects and ocular hypertensive patients (zero RNFL loss
score), but no statistically significant difference was found between the respective
hemifield MD values of patients with ocular hypertension and early glaucoma.
This suggests that in some subjects with ocular hypertension the functional
damage detected by B/Y perimetry may even precede RNFL defects on conversion
to glaucoma
There is a correlation between neural structure, as observed with a confocal
scanning laser tomograph, with RNFL photographs, and with visual function
measured with B/Y perimetry. Although further investigation is needed with a
larger number of subjects to confirm the results of this study, it seems that in many
cases the presence of the B/Y visual field defect may shift the diagnosis from
glaucoma suspect to glaucoma, reveal a more advanced glaucomatous damage
than assessed on the basis of W/W visual field, and therefore improve the
adequacy of our therapeutic decisions.
69
 Short answers to the specific topics on page 30:
1. A statistically highly significant linear correlations were found between
sensitivity values of B/Y visual field and lens transmission index in
nonglaucomatous subjects.
2. The reference level for correcting B/Y perimetry results can be determined more
precisely using fluorometry of the lens than with age alone.
3. The B/Y visual field total and hemifield MDs showed a statistically significant
correlation with respective optic nerve head parameters such as cup shape
measure, rim volume, rim area, mean RNFL thickness and RNFL cross sectional
area.
4. Most of the HRT parameters were better correlated with the B/Y hemifield data
than with the W/W hemifield data. Without the advanced glaucoma subjects the
differences of the correlation coefficients increased.
5. The B/Y visual field mean deviation value showed a statistically significant
correlation with respective diffuse and overall RNFL loss scores.
6. The B/Y hemifield mean deviation combined with the CSM of the Heidelberg
Retina Tomograph showed a good sensitivity (95.2%) and specificity (87.9%)
70
correctly identifying 92.7% of the normal and abnormal hemifields classified by
W/W perimetry.
7. The B/Y visual field term labeled 38%, cup shape measure 52% and RNFL
overall loss score 84% of “normal” W/W hemifields of patients with glaucoma as
abnormal.
71
REFERENCES
Adams AJ, Rodic R, Husted R & Stamper R. (1982) Spectral sensitivity and color
discrimination changes in glaucoma and glaucoma-suspects. Invest Ophthalmol
Vis Sci 23:516-524.
Airaksinen PJ, Burk ROW, Vihanninjoki K, Tuulonen A & Alanko HI. (1995)
Neuroretinal rim volume measurements with the Heidelberg Retina Tomograph.
In:Krieglstein GK (ed.). Glaucoma Update V. Kaden Verlag, p.91-93.
Airaksinen PJ, Drance SM, Douglas GR, Mawson DK & Nieminen H. (1984)
Diffuse and localized nerve fiber loss in glaucoma. Am J Ophthalmol 98:566-571.
Airaksinen PJ, Drance SM, Douglas GR & Schulzer M. (1985) Neuroretinal rim
areas and visual field indices in glaucoma. Am J Ophthalmol 99:107-110.
Airaksinen PJ, Drance SM & Schulzer M. (1985) Neuroretinal rim area in early
glaucoma. Am J Ophthalmol 99:1-4.
Airaksinen PJ & Heijl A. (1983) Visual field and retinal nerve fibre layer in early
glaucoma after optic disc haemorrhage. Acta Ophthalmol (Copenh) 61:186-194.
Airaksinen PJ, Lakowski R, Drance SM & Price M. (1986) Color vision and
retinal nerve fiber layer in early glaucoma. Am J Ophthalmol 101:208-213.
Airaksinen PJ & Nieminen H. (1985) Retinal nerve fiber layer photography in
glaucoma. Ophthalmology 92:877-879.
Airaksinen PJ, Nieminen H & Mustonen E. (1982) Retinal nerve fiber layer
photography with a wide angle fundus camera. Acta Ophthalmol 60:362-368.
Augusteyn RC. (1975) Distribution of fluorescence in the human cataractous lens.
Ophthalmic Res 7:217-224.
72
Balazsi AG, Drance SM, Schulzer M & Douglas GR. (1984) Neuroretinal rim area
in suspected glaucoma and early chronic open-angle glaucoma. Correlation with
parameters of visual function. Arch Ophthalmol 102:1011-1014.
Behrendt T & Wilson LA. (1965) Spectral reflectance photography of the retina.
Am J Ophthalmol 59:1079-1088.
van Best JA, Tjin a tsoi EW, Boot JP & Oosterhuis JA. (1985) In vivo assessment
of lens transmission for blue-green light by autofluorescence measurements.
Ophthalmic Res. 17:90-95.
Bleeker JC, van Best JA, Vrij L, van der Verde EA & Oosterhuis JA. (1986)
Autofluorescence of the lens in diabetic and healthy subjects by fluorophotometry.
Invest Ophthalmol Vis Sci 27:791-794.
Brigatti L & Caprioli J. (1995) Correlation of visual field with scanning confocal
laser optic disc measurements in glaucoma. Arch Ophthalmol 113:1191-1194.
Burk ROW, Rohrschneider K, Völcker HE & Zinser G. (1990) Analysis of threedimensional optic disk topography by laser scanning tomography. Parameter
definition and evaluation of parameter interdependence. In: Nasemann JE, Burk
ROW, eds. Scanning Laser Ophthalmoscopy and Tomography. München:
Quintessenz, p.161-176.
Caprioli J & Miller JM. (1988) Correlation of structure and function in glaucoma.
Quantitative measurements of disc and field. Ophthalmology 95:723-727.
Caprioli J & Miller JM. (1988) Videographic measurements of the optic nerve
topography in glaucoma. Invest Ophthalmol Vis Sci 29:1294-1298.
Chauhan BC, Leblanc RP, McCormick TA & Rogers JB. (1994) Test-retest
variability of topographic measurements with the confocal scanning laser
tomography in patients with glaucoma and control subjects. Am J Ophthalmol
118:9-15.
73
Cioffi GA, Robin AL, Eastman RD, Perell HF, Sarfarazi FA & Kelman SE. (1993)
Confocal laser scanning ophthalmoscope: reproducibility of optic nerve head
topographic measurements with the confocal laser scanning ophthalmoscope.
Ophthalmology 100:57-62.
Delori FC, Gragoudas ES, Francisco R & Pruett RC. (1977) Monochromatic
ophthalmoscopy and fundus photography: The normal fundus. Arch Ophthalmol
95:861-868.
Delori FC & Burns SA. (1996) Fundus reflectance and the measurement of
crystalline lens density. J Opt Soc Am A 13:215-226.
Drance SM. (1969) The early field defects in glaucoma. Invest Ophthalmol Vis Sci
8:84-91.
Drance SM. (1981) Early disturbances of color vision in chronic open angle
glaucoma. Doc Ophthal Proc Ser 26:155-159.
Drance SM, Airaksinen PJ, Price M, Schulzer M, Douglas GR & Tansley BW.
(1986) The correlation of functional and structural measurements in glaucoma
patients and normal subjects. Am J Ophthalmol 102:612-616.
Drance SM, Lakowski R, Schulzer M & Douglas GR. (1981) Acquired color
vision changes in glaucoma: use of 100-Hue Test and Pickford anomaloscope as
predictors of glaucomatous field change. Arch Ophthalmol 99: 829-831.
Dreher AW, Tso PC & Weinreb RN. (1991) Reproducibility of topographic
measurements of the normal and glaucomatous optic nerve head with the laser
tomographic scanner. Am J Ophthalmol 111:221-229.
Dreher AW & Weinreb RN. (1991) Accuracy of topographic measurements in a
model eye with the laser tomographic scanner. Invest Ophthalmol Vis Sci
32:2992-2996.
74
Ducrey NM, Delori FC & Gragoudas ES. (1979) Monochromatic ophthalmoscopy
and fundus photography. II. The Pathological fundus. Arch Ophthalmol 97:288293.
Felius J, de Jong LAMS, van der Berg TJTP & Greve EL. (1995) Functional
characteristics of blue-on-yellow perimetric threshold in glaucoma. Invest
Ophthalmol Vis Sci 36:1665-1674.
Fishman GA, Krill AE & Fishman M. (1974) Acquired color defects in patients
with open-angle glaucoma and ocular hypertension. Mod Probl Ophthalmol
13:335-338.
Frisen L. (1980) Photography of the retinal nerve fiber layer. An optimized
procedure. Br J Ophthalmol 64:641-650.
Foulds WS, Chisholm IA & Bronte-Stewart JM (1974) Effects of raised intraocular pressure on hue discrimination. Mod Probl Ophthalmol 13:328-334.
Funk J, Bornscheuer C & Grehn F. (1988) Neuroretinal rim area and visual field in
glaucoma. Graefes Arch Clin Exp Ophthalmol 226:431-434.
Funk J, Soriano JM & Ebner D. (1993) Correlation between optic disc changes and
visual field loss in patients with unilateral glaucoma. In: Mills RP, ed. Perimetry
Update 1992/93. Amsterdam: Kugler and Ghedini p.159-163.
Guthauser U, Flammer J & Niesel P. (1987) The relationship between the visual
field and the optic nerve head in glaucomas. Graefes Arch Clin Exp Ophthalmol
225:129-132.
Grützner P & Schleicher S. (1972) Acquired color vision defects in glaucoma
patients. Mod Probl Ophthalmol 11:136-140
Hart WM Jr & Gordon MO. (1984) Color perimetry of glaucomatous visual field
defects. Ophthalmology 91:338-346.
75
Hart WM Jr, Hartz RK, Hagen RW & Clark KW. (1984) Color contrast perimetry.
Invest Ophthalmol Vis Sci 25:400-413.
Hart WM, Silverman SE, Trick GL, Nesher R & Gordon MO. (1990)
Glaucomatous visual field damage: luminance and color contrast sensitivities.
Invest Ophthalmol Vis Sci 31:359-367.
Hart WM Jr, Yablonski M, Kass MA & Becker B. (1978) Quantitative visual field
and optic disc correlates early in glaucoma. Arch Ophthalmol 96:2209-2211.
Heijl A, Lindgren G & Olsson J. (1987) Normal variability of static perimetric
threshold values across the central visual field. Arch Ophthalmol 105:1544-1549.
Horn FK, Jonas JB, Korth M, Jünemann A & Gründler A. (1997) The full-field
flicker test in early diagnosis of chronic open-angle glaucoma. Am J Ophthalmol
123:313-319.
Hoyt WF, Frisèn LL & Newman NM. (1973) Fundoscopy of nerve fiber layer
defects in glaucoma. Invest Ophthalmol 12:814-829.
Iester M, Mikelberg FS, Courtright P & Drance SM. (1997) Correlation between
the visual field indices and Heidelberg Retina Tomograph parameters. J Glaucoma
6:78-82.
Iwata K, Nanba K & Abe H. (1981) Typical slit-like retinal nerve fiber layer
defect and corresponding scotoma. Acta Soc Ophthalmol Jpn 85:1791-1803.
Jacobs R & Krohn DL. (1981) Fluorescence intensity profile of human lens
sections. Invest Ophthalmol Vis Sci 22:117-120.
de Jong LAMS, Felius J, van den Berg TJTP & Greve EL. (1992) Use of a blue
filter in visual field analysis. Br J Ophthalmol 76:447-448.
Johnson CA. (1994) Selective versus nonselective losses in glaucoma. J Glaucoma
3:S32-S44.
76
Johnson CA. (1996) Diagnostic value of short-wavelength automated perimetry.
Current Opinion in Ophthalmology 7:54-58.
Johnson CA, Adams AJ, Casson EJ & Brandt JD. (1993) Progression of early
glaucomatous visual field loss as detected by blue-on-yellow and standard whiteon-white perimetry. Arch Ophthalmol 111:651-656.
Johnson CA, Adams AJ, Casson EJ & Brandt JD. (1993) Blue-on-yellow
perimetry can predict the development of glaucomatous visual field loss. Arch
Ophthalmol 111:645-650.
Johnson CA, Adams AJ & Lewis RA. (1989) Evidence for a neural basis of agerelated visual field loss in normal observer. Invest Ophthalmol Vis Sci 30:20562064.
Johnson CA, Adams AJ & Lewis RA. (1989) Patterns of early glaucomatous
visual field loss for short wavelength sensitive pathways. ARVO Abstracts. Invest
Ophthalmol Vis Sci 30:S56.
Johnson CA, Adams AJ, Twelker JD & Quigg JM. (1988) Age-related changes in
the central visual field for short-wavelength-sensitive pathways. J Opt Soc Am A
5:2131-2139.
Johnson CA Brandt JD, Khong AM & Adams AJ. (1995) Short-wavelength
automated perimetry in low-, medium-, and high-risk ocular hypertensive eyes.
Arch Ophthalmol 113:70-76.
Johnson CA, Howard DL, Marshall D & Shu H. (1993) A noninvasive videobased method for measuring lens transmission properties of the human eye. Optom
Vis Sci 70:944-955.
Jonas JB, Gusek GC & Naumann GOH. (1988) Optic disc morphometry in chronic
primary open-angle glaucoma I. Morphometric intrapapillary characteristics.
Graefe’s Arch Clin Exp Ophthalmol 226:522-530.
77
Jonas JB, Gusek GC & Naumann GO. (1988) Optic disc morphometry in chronic
primary open-angle glaucoma. II. Correlation of the intrapapillary morphometric
data to visual field indices. Graefes Arch Clin Exp Ophthalmol 226:531-538.
Jonas JB & Naumann GOH. (1991) The anatomic structure of the normal and
glaucomatous optic nerve. In: Kriegelstein GK (ed.). Glaucoma Update IV. Berlin:
Springer-Verlag, p.66-73.
King-Smith PE, Lubow M & Benes SC. (1984) Selective damage to chromatic
mechanisms in neuro-ophthalmic diseases I. Review of published evidence. Doc
Ophthalmol 58:241-250.
Kitahara K, Tamaki R, Noji J, Kandatsu A & Matsuzaki H. (1982) Extrafoveal
Stiles pi mechanisms. Doc Ophthalmol 35:397-404.
Kruse FE, Burk RO, Volcker HE, Zinser G & Harbarth U. (1989) Reproducibility
of topographic measurements of the optic nerve head with laser tomographic
scanning. Ophthalmology 96:1320-1324.
Lakowski R, Bryett J & Drance SM. (1972) A study of colour vision in ocular
hypertensives. Can J Ophthalmol 7:86-95.
Lakowski R & Drance SM. (1979) Acquired dyschromatopsias. The earliest
functional losses in glaucoma. Doc Ophthalmol Proc Ser 19:159-165.
Lee BB, Martin PR & Valberg A. (1989) Sensitivity of macaque ganglion cells to
luminance and chromatic flicker. J Physiol (Lond) 414:223-243.
Lerman S & Borkman R. (1976) Spectroscopic evaluation and classification of the
normal, aging and cataractous lens. Ophthalmic Res 8:335-353.
Lewis RA, Johnson CA & Adams AJ. (1993) Automated perimetry and short
wavelength sensitivity in patients with asymmetric intraocular pressures. Graefe’s
Arch Clin Exp Ophthalmol 231:274-278.
78
Lusky M, Bosem ME & Weinreb RN. (1993) Reproducibility of the optic nerve
head topography measurements in eyes with undilated pupils. J Glaucoma 2:104109.
Marre M. (1972) Clinical examination of the three color vision mechanisms in
acquired color vision defects. Mod Probl Ophthalmol 11:224-227.
Merigan WH & Maunsell JHR. (1993) How parallel are the primate visual
pathways? Ann Rev Neurosci 16:369-402.
Mikelberg FS, Wijsman K & Schulzer M. (1993) Reproducibility of topographic
parameters obtained with the Heidelberg Retina Tomograph. J Glaucoma 2:101103.
Miller NR & George T. (1978) Monochromatic (red-free) photography and
ophthalmoscopy of the peripapillary retinal nerve fiber layer. Invest Ophthalmol
Vis Sci 17:1121-1124.
Moss ID, Wild JM & Whitaker DJ. (1995) The influence of age-related cataract on
blue-on-yellow perimetry. Invest Ophthalmol Vis Sci 36:764-773.
Niessen AGJE, Van den Berg TJTP, Langerhorst CT & Bossuyt PM. (1995)
Grading of retinal nerve fiber layer with a photographic reference set. Am J
Ophthalmol 120:577-586.
Norren D van & Vos JJ. (1974) Spectral transmission of the human ocular media.
Vis Res 14:1237-1244.
Occhipinti JR, Mosier MA & Burnstein NL. (1986) Autofluorescence and light
transmission in the aging crystalline lens. Ophthalmologica, Basel 192:203-209.
Pederson JE & Anderson DR (1980) The mode of progressive disc cupping in
ocular hypertension and glaucoma. Arch Ophthalmol 98:490-495.
Peli E, Hedges TR, McInnes T, Hamlin J & Schwartz B. (1987) Nerve fiber layer
photography: A comparative study. Acta Ophthalmol (Copenh) 65:71-80.
79
Peli E, Hedges TR & Schwartz B. (1986) Computerized enhancement of retinal
nerve fiber layer. Acta Ophthalmol (Copenh) 64:113-122.
Pokorny J, Smith VC & Lutze M. (1987) Aging of the human lens. Appl Opt
26:1437-1440.
Quigley HA & Addicks EM. (1982) Quantitative studies of the retinal nerve fiber
layer defects. Arch Ophthalmol 100:807-814.
Quigley HA, Dunkelberger GR & Green WR. (1988) Chronic human glaucoma
causing selectively greater loss of large optic nerve fibers. Ophthalmology 95:357363.
Quigley HA, Dunkelberg GR & Green WR. (1989) Retinal ganglion cell atrophy
correlated with automated perimetry in human eyes with glaucoma. Am J
Ophthalmol 107:453-464.
Quigley HA, Katz J, Derick RJ, Gilbert D & Sommer A. (1992) An evaluation of
optic disc and nerve fiber layer examinations in monitoring progression of early
glaucoma damage. Ophthalmology 99:19-28.
Quigley HA, Reacher M, Katz J, Strahlman E, Gilbert D & Scott R. (1993)
Quantitative grading of the nerve fiber layer photographs. Ophthalmology
100:1800-1807.
Quigley HA, Sanches RM, Dunkelberger GR, Henaut NL & Baginski TA. (1987)
Chronic glaucoma selectively damages large optic nerve fibers. Invest Ophthalmol
Vis Sci 28:913-918.
Rohrschneider K, Burk ROW, Kruse FE & Völcker HE. (1994) Reproducibility of
the optic nerve head topography with a new laser tomographic scanning device.
Ophthalmology 101:1044-1049.
80
Rohrschneider K, Burk ROW & Völcker HE. (1993) Reproducibility of
topographic data acquisition in normal and glaucomatous optic nerve head with
the laser tomographic scanner. Graefe’s Arch Clin Exp Ophthalmol 231:457-464.
Sample PA, Cook JN & Weinreb RN. (1993) Variability and sensitivity of shortwavelength color visual field in normal and glaucoma eyes. In: Noninvasive
Assessment of the Visual System Tech Digest Ser p.292-295.
Sample PA, Esterson FD & Weinreb RN. (1989) A practical method for obtaining
an index of lens density with an automated perimeter. Invest Ophthalmol Vis Sci
30:786-787.
Sample PA, Esterson FD, Weireb RN & Boynton RM. (1988) The aging lens: in
vivo assessment of light absorption in 84 human eyes. Invest Ophthalmol Vis Sci
29:1306-1311.
Sample PA, Irak I, Martinez GA & Yamagishi N. (1997) Asymmetries in the
normal short-wavelength visual field: implications for short-wavelength
automated perimetry. Am J Ophthalmol 124:46-52.
Sample PA, Johnson CA, Haegerstrom-Portnoy G & Adams AJ. (1996) Optimum
parameters for short-wavelength automated perimetry. J Glaucoma 5:375-383.
Sample PA, Juang PSC & Weinreb RN. (1994) Short-wavelength automated
perimetry for analysis of secondary and normal tension glaucoma. ARVO
Abstracts. Invest Ophthalmol Vis Sci 35:2189.
Sample PA, Martinez GA & Weinreb RN. (1993) Color visual fields: a 5-year
prospective study in eyes with primary open angle glaucoma. In: Mills RP, ed.
Perimetry Update 1992/93. Amsterdam: Kugler and Ghedini p.473-476.
Sample PA, Martinez GA & Weinreb RN. (1994) Short-wavelength automated
perimetry without lens density testing. Am J Ophthalmol 118:632-641.
81
Sample PA, Plummer DJ & Weinreb RN. (1995) Standard achromatic automated
perimetry vs. short-wavelength automated perimetry for following glaucoma. In:
Kriglestein GK, ed. Glaucoma update V. Berlin: Springler-Verlag p.197-204.
Sample PA, Taylor JDN, Martinez GA, Lusky M & Weinreb RN. (1993) Shortwavelength color visual field in glaucoma suspects at risk. Am J Ophthalmol
115:225-233.
Sample PA & Weinreb RN. (1990) Color perimetry for assessment of primary
open angle glaucoma. Invest Ophthalmol Vis Sci 31:1869-1875.
Sample PA & Weinreb RN. (1992) Progressive color visual field loss in glaucoma.
Invest Ophthalmol Vis Sci 33:2068-2071.
Sample PA, Weinreb RN & Boynton RM. (1986) Blue-on-yellow color perimetry.
ARVO Abstracts. Invest Ophthalmol Vis Sci 27:159.
Sample PA, Weinreb RN & Boynton RM. (1986) Acquired dyschromatopsia in
glaucoma. Surv Ophthalmol 31:54-64.
Siik S, Airaksinen PJ, Tuulonen A, Alanko H & Nieminen H. (1991) Lens
autofluorescence in healthy individuals. Acta Ophthalmol (Copenh) 69:187-192.
Siik S, Airaksinen PJ, Tuulonen A & Nieminen H. (1993) Autofluorescence in
cataractous human lens and its relationship to light scatter. Acta Ophthalmol
(Copenh) 71:388-392.
Sommer A, D'Anna SA, Kues HA & George T. (1983) High-resolution
photography of the retinal nerve fiber layer. Am J Ophthalmol 96:535-539.
Sommer A, Katz J, Quigley HA, Miller NR, Robin AL, Richter RC & Witt KA.
(1991) Clinically detectable nerve fiber atrophy precedes the onset of
glaucomatous field loss. Arch Ophthalmol 109:77-83.
Stiles WS. (1959) Color vision: the approach through increment threshold
sensitivity. Proc Natl Acad Sci U S A 45:100-114.
82
Tsai CS, Zangwill L, Sample PA, Garden V, Bartsch DU & Weinreb RN. (1995)
Correlation of peripapillary retinal height and visual field in glaucoma and normal
subjects. J Glaucoma 4:110-116.
Tuulonen A & Airaksinen PJ. (1991) Initial glaucomatous optic disc and retinal
nerve fiber abnormalities and their progression. Am J Ophthalmol 111:485-490.
Tuulonen A, Lehtola J & Airaksinen PJ. (1993) Nerve fiber layer defects with
normal visual fields. Ophthalmology 100:587-598.
Tuulonen A, Vihanninjoki K, Airaksinen PJ, Alanko H & Nieminen H. (1994) The
effect of reference levels on neuroretinal rim area and rim volume measurements
in Heidelberg Retina Tomograph (HRT). ARVO Abstracts. Invest Ophthalmol Vis
Sci 35(4):1729.
Vogt A. (1917) Die Nervenfaserstreifung der menschlichen Netzhaut mit
besonderer Berücksichtigung der Differentialdiagnose gegenüber pathologischen
streifenförmigen Reflexen (präretinalen Fältelungen). Klin Monatsbl Augenheilkd
58:399-411.
Weinreb RN, Dreher AW & Bille J. (1989) Quantitative assessment of the optic
nerve head with the laser tomographic scanner. Int Ophthalmol 13:25-27.
Weinreb RN, Shakiba S, Sample PA, Shahrokni S, van Horn S, Garden VS,
Asawaphureekorn S & Zangwill L. (1995) Association between quantitative nerve
fiber layer measurement and visual field loss in glaucoma. Am J Ophthalmol
120:732-738.
Wild JM, Moss ID, Whitaker D & O’Neill EC. (1995) The statistical interpretation
of blue-on-yellow visual field loss. Invest Ophthalmol Vis Sci 36:1398-1410.
Wild JM & Hudson C. (1995) The attenuation of blue-on-yellow perimetry by the
macular pigment. Ophthalmology 102:911-917.
83
Yappert MC, Lal S & Borchman D. (1992) Age dependence and distribution of
green and blue fluorophores in human lens homogenates. Invest Ophthalmol Vis
Sci 33:3555-3560.
Yeh T & Smith VC. (1989) The effect of the background luminance on the cone
sensitivity functions. Invest Ophthalmol Vis Sci 30:2077-2086.
Zeimer RC & Noth JM. (1984) A new method of measuring in vivo the lens
transmittance and study of lens scatter, fluorescence and transmittance.
Ophthalmic Res 16:246-255.
Zeyen TG & Caprioli J. (1993) Progression of disc and field damage in early
glaucoma. Arch Ophthalmol 111:62-65.
Zinser G, Wijnaendts-van Resandt RW, Dreher AW, Weinreb RN, Harbarth U &
Schröder H. (1989) Confocal laser tomographic scanning of the eye. In: Burk RO,
ed., Transactions SPIE. 1161:337-344.
Zisman F, King-Smith PE & Bhargava SK. (1978) Spectral sensitivities of
acquired color defects analyzed in terms of color opponent theory. Mod Probl
Ophthalmol 19:254-257.
84
FIGURE LEGENDS AND FIGURES
Fig. 1. The correlation of the blue-on-yellow visual field mean sensitivity values
(program 24-2) in non-glaucomatous individuals plotted against the lens
transmission index. There is a statistically highly significant correlation (R = 0.83;
P < 0.0001). The 95 % prediction interval is shown.
Fig. 2. Scatterplot of the association of the white-on-white (W/W) visual field
mean deviation (MD) values with the lens- and age-adjusted blue-on-yellow (B/Y)
visual field MD values in the five clinical groups.
Fig. 3. Scatterplot of the association of the white-on-white (W/W) visual field
mean deviation (MD) values with the cup shape measure in the five clinical
groups.
Fig. 4. Scatterplot of the association of the lens- and age-adjusted blue-on-yellow
(B/Y) visual field mean deviation (MD) values with the cup shape measure in the
five clinical groups.
Fig. 5. Scatterplot of the association of the lens- and age-adjusted blue-on-yellow
(B/Y) visual field inferior hemifield mean deviation (HMD) values with the cup
shape measure obtained from the superior part of the optic disc in the five clinical
groups.
85
Fig. 6. The age- and lens-adjusted mean deviation (MD) values of the blue-onyellow (B/Y) visual field with diffuse score of the retinal nerve fiber layer (RNFL)
loss in the five clinical groups (Pearson´s R = - 0.73; P< 0.0001).
Fig. 7. The age- and lens-adjusted hemifield mean deviation (HMD) of white-onwhite (W/W) visual field was better than - 2 dB in 55 hemifields of normal
subjects (N=32), in 12 hemifields of patients with ocular hypertension (N=6) and
in 19 hemifields of patients with early glaucoma (N=12). Scatterplot shows the
relationship between the HMD of these W/W visual fields and the HMD of the
respective blue-on-yellow (B/Y) visual fields. The HMD of the B/Y visual fields
in the normals (0.39dB ± 2.95) was statistically significantly different (ANOVA
P= 0.0002) both from the ocular hypertensives (-2.52dB ± 4.34) and the early
glaucoma patients (-3.56dB ± 4.99). However, the HMD of the B/Y visual fields
in the ocular hypertensives did not differ from that of the patients with early
glaucoma.
Fig. 8. The age- and lens-adjusted hemifield mean deviation (HMD) of blue-onyellow (B/Y) visual fields obtained from “normal” (HMD better than - 2 dB)
white-on-white hemifields, plotted against the overall score of RNFL loss from the
respective hemisphere in patients with early glaucoma (Pearson´s R = - 0.56; P =
0.012) and ocular hypertension.
Fig. 9. The relationship of the blue-on-yellow visual field mean sensitivity values
to the lens transmission index in the ocular hypertensive groups with the 95%
86
prediction interval of the normals. The circles represent ocular hypertensive
patients who had an IOP of greater than 22 mmHg with normal optic disc, normal
retinal nerve fiber layer (RNFL) and normal Humphrey 30-2 visual fields. The
triangles represent ocular hypertensive patients with normal Humphrey 30-2 visual
fields but abnormal RNFL or optic disc.
Fig. 10. The relationship of the blue-on-yellow visual field mean sensitivity values
to the lens transmission index in the different groups of patients with glaucoma.
Circles represent early (MD < -5 dB), triangles moderate (-5 > MD > -10 dB) and
squares patients with advanced glaucoma (MD < -10 dB) (tested with Humphrey
30-2 white-on-white perimetry). The 95% prediction interval of the normals is
superimposed on the data points.
Fig. 11. The lens transmission values showed a good correlation to age (R= 0.80;
P<0.0001). The interindividual variation of lens transmission properties increases
with age.
Fig. 12. Scattergram showing the relationship between the blue-on-yellow visual
field mean deviation values in patients with glaucoma adjusted for their age and
lens transmission index. The deviations from the linear line show differences
between two ways of correction.