An Evaluation of Optic Disc

8/10/2019 An Evaluation of Optic Disc

1/10

n

Evaluation of Optic Disc and

erve

Fiber Layer Examinations

in Monitoring Progression

of

Early Glaucoma Damage

Harry A Quigley MD, Joanne Katz MS Robert]. Derick MD,

Donna Gilbert Alfred Sommer MD

F

rom annual examinations

of

813 ocular hypertensive eyes, the authors compared

optic disc and nerve fiber layer photographs in 2 age-matched subgroups: 37 eyes

that converted to abnormal visual field tests at the end

of

a 5-year period and 37 control

eyes that retained normal field tests. Disc change was detected in only 7

of

37 (19 )

converters

to

field loss and in 1 of 37 (3 ) controls. Progressive nerve fiber layer

atrophy was observed in 8

of

37 (49 ) converters and in 3

of

37 (8 ) controls. Serial

nerve fiber layer examination was more sensitive than color disc evaluation in the de

tection of progressive glaucoma damage at this early stage of glaucoma. The evaluation

of cup-to-disc ratio or

of

the nerve fiber layer appearance in the initial photograph taken

5 years before field loss were equally predictive of future field damage. The position

of

nerve fiber layer defects was highly correlated with the location of subsequent visual

field loss.

phthalmology

1992; 99:19 28

Since glaucoma damage is largely irreversible, it is im

perative to predict accurately those eyes at greatest risk

of

future injury. For clinical trials

l

2

and routine clinical

management, surveillance parameters are needed to

monitor the health of the optic nerve. The most useful

indication

of

definitive optic nerve damage

is

abnormality

in visual field testing. The aim

of

management in glau

coma suspects is to predict and, it is hoped, to avoid the

development

of

field loss.

Originally received: July 19 1991.

Revision accepted: September 13 1991.

From the Dana Center for Preventive Ophthalmology and the Glaucoma

Service, Wilmer Institute at Johns Hopkins School of Medicine and the

Johns Hopkins School of Public Health and Hygiene, Baltimore.

Supported in part by Public Health Service Research Grants EY 02120,

EY

01765,

EY

03605,

RR

04060, and

by

funds provided by National

Glaucoma Research, a program of American Health Assistance Foun

dation, Rockville, Maryland, and by a Senior Investigator Award from

Research to Prevent Blindness, Inc, New York, New York.

Reprint requests to Harry A. Quigley, MD, Wilmer 120, Dana Center

for Preventive Ophthalmology, Johns Hopkins Hospital, 600 N Wolfe

St, Baltimore,

MD

21205.

Among eyes that are suspected

of

glaucoma because

of

elevated intraocular pressure lOP) or positive family

history, only a small percent develop field loss per year

of

observation.

1 8

The risk factors for field loss include:

higher

lOP

level, older age and black race.

I

I

Probable

additional risk factors are myopia, diabetes, family history

of

glaucoma, and hypertension.

9

11

The risk factor most

strongly associated with later field loss is larger cup-to

disc ratio. This probably reflects greater loss

of

nerve fibers

before initial examination.

12

To detect damage before the stage offield loss, we ex

amine the optic disc

l3

-

16

and, more recently, the nerve

fiber layer. 17-21 The disc features that have been studied

include the cup-to-disc ratio,

13

the width

of

he disc rim, 14

and the neural rim area.

15

16

These are surrogate estimates

of

the amount

of

nerve tissue present. Cross-sectional

studies have shown that these parameters correlate with

the likelihood

of

existing field loss, but do a poor job

of

separating normal from suspect and field-damaged eyes.

Few prospective studies

of

the predictive power

of

initial

or serial disc examination have been performed. Newer

techniques for image analysis of the disc have not been

tested prospectively.

19


2/10

Ophthalmology Volume 99, Number 1, January 992

Ten years ago, we initiated a longitudinal follow-up of

more than 800 ocular hypertensive glaucoma suspects,

along with additional normal and field-damaged eyes.

20

A recent report demonstrates the considerable predictive

power of nerve fiber layer examination in these eyes.

2

Here, we present a case-control study of the predictive

value

of

optic disc and nerve fiber layer data from a se-

lected group of these ocular hypertensive subjects, some

of whom developed field loss 5 years after their initial

examination (cases) and some of whom did not (controls).

Our primary aim was to compare the value of serial disc

examination with serial nerve fiber layer examination.

aterials and ethods

The Nerve Fiber Layer Study has been conducted at our

institution since 1981.

20

In

brief, we recruited by adver

tisement and referral 341 persons with reproducibly nor

mal visual field tests (as defined in a detailed protocol on

the Goldmann perimeter

o

) and no family history of glau

coma (visually healthy group), 813 ocular hypertensive

persons with lOP greater than 21 mmHg with 2 normal

visual field tests (the same criteria

as

for visually healthy

subjects), and 99 persons with defined, reproducible visual

field defects on the Goldmann perimeter. Field defects

were defined as one or more of

the following: (1) a nasal

step 10 in extent to 2 isopters or in 1 isopter when there

is an associated scotoma present in the same hemifield

that is 0. 2 log units in depth; (2) a paracentral scotoma

at least 5 in diameter and 0.4 log unit depressed com

pared with the surrounding isopter; and (3) an arcuate

extension of the blind spot of at least 45 radial extent,

0.3 log units deep. Each subject underwent annual

ophthalmic examination including color stereoscopic

photography of the optic disc and red-free photography

of

the retinal nerve fiber layer. Since 1985, study subjects

have been tested yearly with the 30-2 program of the

Humphrey Field Analyzer; however, for consistency, field

loss was defined by defects on the Goldmann instrument.

Among the ocular hypertensive eyes, 93 eyes

of80

per

sons have developed a reproducible defect on the Gold

mann field test. For this report, we include 37 of these

eyes that have converted to

field

loss in whom photographs

are available in each

of

the 5 years before conversion.

Because four persons developed field loss in both eyes,

this represents

37

eyes of

33

persons. The analyses to be

presented here do not differ substantively when only one

eye

of

each bilateral converter is included. In the remain

der of the 93 converting eyes, perimetric loss occurred

after a shorter period of follow-up. We randomly selected

37

eyes of

37

ocular hypertensive persons in whom pho

tographs were available for a similar 5-year period of ob

servation, but whose field tests have remained within our

criteria of normality. These ocular hypertensive controls

were matched for race, sex, and

age

to within 3 years,

with the

33

converters (Table

1). In

each group, approx

imately 60 of subjects were white, 40 black, and 40

male. We include similar data from 88 eyes

of

88 visually

healthy persons who were group-matched for

age

and race

to converters and controls. These visually healthy subjects

were recruited from local church groups and from other

studies, including the population-based Baltimore Eye

Survey.

One observer evaluated color stereoscopic and nerve

fiber layer photographs

of

converter, ocular hypertensive

control, and their fellow eyes without knowledge of patient

status. During color disc examination, the immediate re

gion around the disc containing the nerve fiber layer was

visible to the observer. For nerve fiber layer evaluation,

the optic disc was covered. Color and nerve fiber layer

photographs were examined in separate sessions.

For color disc examination, the vertical and horizontal

diameter of the optic disc and of the cup were measured

with a micrometer overlay placed within the stereoviewing

device. The micrometer scale measures an object the size

of he disc to

an

accuracy

of0.05

mm (allowing the cup

to-disc ratio to be calculated to 2 significant figures). Cup

dimensions were measured by contour of he stereoscopic

image, although it is

cl

e

ar

that color clues play a role in

this judgment. The cup margin was taken as the point of

maximum slope change, or, in gradually sloping cup

margins, apoint halfway down the slope. The coefficients

of

variation

of

repeated measures for disc and cup size

were less than 2 . The cup-to-disc ratio was calculated

from these data. We also calculated disc diameter in mil

limeters, accounting for image magnification with a for

mula based on the spherical equivalent refractive error.

After measurement of cup and disc diameters in each

photograph, the observer compared photographs of the

same disc at different points in time from the initial with

the final ones in a specially designed viewer. This instru

ment allowed simultaneous comparison of two stereopairs

through one binocular eyepiece. There were either 4 or 5

stereopairs of each eye, taken at annual intervals. Any

apparent enlargement in cup size, or any change in rim

slope or vessel position was labeled disc progression. The

observer knew the actual temporal sequence

of

photog

raphy. These data most closely approach the clinical set

ting and are reported in detail here. This determination

Table 1 Patient

Characteristics

No. of

No.

of

Mean

Age (95

Confidence

Persons Eyes

(yrs)

Interval)

Converters

33

37 66 63 , 69)

Ocula r hypertensive controls

37 37

67 (64,71)

Normals

88 88

65 (64,66)

20


3/10

Quigley et al . Progression of Early Glaucoma Damage

Table 2. Vertical Disc Diameter Data

Vertical

Diameter mm)

Initial Diameter

hange

in

Diameter

Converters

Controls

CI = confidence interval.

No.

37

37

Mean

1.72

1.69

(95 CI)

Mean

(95

CI)

(1.65, 1.79) -0.010 (-0.023,0.003)

(1.63, 1.75)

-0.013

(-0.035, 0.009)

Horizontal disc diameter data are similar. Data are corrected for magnification.

22

was performed independently from the measurement

of

cup and disc diameters. A second evaluation was per

formed

of

photographs

of

all subjects in which the observer

was masked

to

time sequence.

In addition

to the

comparisons

and

measurements

above,

in

each color photograph,

the

observer

noted the

presence or absence of either disc hemorrhage or discon

tinuities

in the

pigmentation of

the

immediate peripapil

lary area (crescents). A crescent was defined as either hy

popigmentation or hyperpigmentation adjacent

to

the disc

(alpha-type) or increased visibility

of

the choroidal vessels

(beta-type).23 The flange of sclera separating

the

optic

nerve from the choroid was

not

considered as crescent.

Nerve fiber layer photographs were evaluated with a ,

hand magnifier

and

graded as either normal, unreadable,

or as having wedge-shaped

or

diffuse atrophlo,21,24

of

the

upper

or

lower nerve fiber layer. The reproducibility of

this technique has been reported.

20

21

In addition, the ob

server designated atrophy as either mild or severe. Mild

defects were those

that

were

just

beyond

the

range

of

nor

mal variation, while severe defects were unequivocal

and

obviously different from normal . Progressive nerve fiber

layer change was indicated either by conversion from

normal to either defect type, or by change from mild to

severe atrophy.

Our

study is not a

treatment

trial. Seventy-six percent

of converters (28

of

37)

and

51

of

ocular hypertensive

controls (19

of

37) were being prescribed treatment to

lower their lOP at some point during the time period

of

this study. The referring ophthalmologist

made the

de

cision whether to treat.

esults

Baseline

Optic

Disc

and

Cup-to-disc Ratio Data

The initial

measurement

of mean vertical

and

horizontal

disc diameter in converters

and

ocular hypertensive con

trols did not differ significantly (Table 2).

Nor

was there

any significant change in disc diameter from initial to

final photographs in either conver ters or ocular hyperten

sive

controls. The ratio of vertical

to

horizontal disc di

ameter was also

not

significantly different among con

verters, ocular hypertens ive controls,

and

visually healthy

subjects.

The mean initial cup-to-disc ratio

of

converters was

significantly larger

than

that

of

ocular hypertensive con

trols

and

visually healthy subjects both vertically and hor

izontally (Table 3). Ocular hypertensive controls

had

sig

nificantly larger mean cup-to-disc ratios than visually

healthy subjects.

An

initial cup-to-disc ratio;;:: 0.55 pro

vided the best balance

of

sensitivity and specificity in dif

ferentiating converters from ocular hypertensive controls.

With

this criterion, sensitivity was

59

(22 of 37 eyes)

and specificity was 73 (i.e., 27 of 37 ocular hypertensive

control eyes had a cup-to-disc ratio


4/10

Ophthalmology olume

99, Number 1, January 992

had mean cup-to-disc ratios that were larger vertically than

horizontally, without major group differences (Table

3).

Cup-to-disc ratio

is,

of course, a function of the di

ameter

of

the optic disc.

25

A detailed analysis

of

adjusting

for disc diameter in the predictive value

of

these data

is

in preparation.

Asymmetry between the cup-to-disc ratio in the con

verting eye and its fellow eye at baseline was an ineffective

criterion for separating converters from ocular hyperten

sive controls. The fellow eyes of seven converters already

had

field

loss at the initial visit, hence these pairs were

removed from this analysis. A criterion of

O . l

cup-to

disc ratio difference between eyes had a sensitivity

of 39

and specificity

of 54 ,

while for an asymmetric difference

of 0 . 2

units it was 10 and

83 ,

respectively.

Change in

Optic Disc Appearance

The observer judged that a qualitative worsening in disc

appearance had occurred from the initial to the final pho

tographs in 7

of

37 converter eyes (19 ) and in I

of

37

ocular hypertensive controls

(3 ).

The change was de

'tected at a mean time before field loss detection of 1.6

years (1.0 [standard deviation]). The measured cup-to

disc ratio increased in each

of

these 7 progressive, con

verter eyes either vertically, horizontally, or both by more

than 0.05 units. The mean increase for these 7 eyes was

0.13 units, significantly greater than the mean increase

for the remaining 30 converters of 0.006 units

t

test,

P

< 0.001). The ocular hypertensive controls had neither

no significant mean increase in vertical cup-to-disc ratio,

nor

was

the mean change in horizontal cup-to-disc ratio sig-

nificant either in all converters or in all controls (Table

3).

When the observerwas masked as to temporal sequence

during evaluation of color photographs, the rates

of

pro

gression were essentially unchanged from those of he first,

temporally unmasked reading

(6

of 37

for converters

[16 ]

and 1

of 37

for controls

[3 ]).

One converter eye

in the masked reading was judged progressive that was

considered unchanged in the unmasked reading, while

two eyes judged progressive in the unmasked reading were

not so selected under masked conditions. These three eyes

that differed in the two series

of

readings had the smallest

measured changes in cup-to-disc ratio of the nine eyes in

both series that were judged progressive. In no case was

an earlier photograph labeled worse than one later in tem

poral sequence.

Converters

and

ocular hypertensive controls did not

differ in the number that changed in either vertical or

horizontal cup-to-disc ratio using categorical separations.

For example, with a criterion for vertical change

f ~ 0 . 0 5

cup-to-disc ratio unit (e.g., from 0.50 to 0.55), the pro

portion of converters and controls that had an enlarge

ment of this size was nearly identical (8 of

37

and 9 of

37, respectively; Table 4).

t

is worth noting that 6 of the

8 converter eyes that had

~ 0 . 0 5

unit increase vertically

were also judged to have progressed (this is 6 of the 7

qualitatively progressive converter

eyes).

There were 4 eyes

with >0.1 unit change, and all were converters judged to

have progressed.

22

Table

4.

Change in Cup-to-disc Ratio Data

Cup-to-Disc

Change (units)

Vertical

+0.05

-0.05

+0.10

-0.10

Horizontal

+0.05

-0.05

+0.10

-0.10

Converters

8

(22 )

2

(5 )

4(11 )

o

11 (30 )

3(8 )

4(11 )

o

Ocular Hypertensive

Controls

9(24 )

3 (8 )

1

(3 )

o

6 (16 )

3 (8 )

o

o

All eyes

with

+0.10 increase in cup-to-disc ratio were among those

judged

to be

progressive subjectively.

We compared the change in cup-to-disc ratio in the

two principal meridians (e.g., vertical exceeding horizontal

cup-to-disc ratio change by 0.05 unit). In both converters

and controls, there was no preponderance of vertical cup

enlargements over horizontal enlargements in such anal

yses regardless

of

he criterion selected. Among converters,

the change in vertical cup-to-disc exceeded that in the

horizontal meridian by ~ 0 . 0 5 units in 6 of 37 eyes, while

an equal number

(6

of 37) had a horizontal change that

exceeded the vertical change by the same criterion. The

control group had 10 of

37

with more vertical than hor

izontal change

~ 0 . 0 5

units), and 7

of 7

with horizontal

exceeding vertical change.

We compared the measured vertical cup enlargement

in eyes that were subjectively judged to have a progressive

change

(7

eyes in the converter group and 1 eye in the

control group) with the measurements in the remaining

66 eyes that did not change qualitatively. A 0.05 unit

enlargement occurred in 7

of

8 progressive eyes

(88 );

only 10 of 66 nonprogressive control eyes or converters

changed by this amount (specificity,

85 ).

An increase in

the criterion

of

change to

0.1

unit (in vertical ratio) still

included 5

of8

progressive eyes (63 ) and was

100

spe

cific (none

of

the nonprogressive eyes changed by this

amount).

Nerve

Fiber

Layer

Examination

Data

. The initial or final photographs of the nerve fiber layer

were readable in

35

of

37

converters and in 34 of

37

con

trols (Table

5).

The initial nerve fiber layer reading was

abnormal in 20

of 37

converters

(54 of

the total or

57

of those with readable photographs). Controls had initial

nerve fiber layer abnormality in

12 of 37

eyes

(32 of

the

total or

35 of

those with readable photographs). These

are similar to the rates previously reported in this popu

lation.

21

A worsening in nerve fiber layer appearance was

detected in

18 of 7

converters

(49 )

and in 2

of 7

con

trols

(5 )

(Figs 1 and 2). The nerve fiber layer change was

detected a mean of 1.4 years before field loss (

1.0

[stan-


5/10

uigleyet al . Progression of Early Glaucoma Damage

Table 5. Nerve Fiber Layer Data

Converter Control P

Value

Normal throughout

22 (8)

60 (22)

0.002

Initially abnormal

54

(20)

32 (12)

0.10

Defect developed

49

(18)

5 (2)


6/10

Ophthalmology

Volume 99, Number

1,

January 992

Figure 2.

This

subject had a

norm

al nerve fiber layer initially

upper

left),

then

in the first year

of

follow-up developed a local nerve fiber layer defect

(lower left, arrows). This broadened into diffuse atrophy

upper right, third

year;

lower

right, fourth year). The conversion to abnormal field testing

was at

the

time of

the

fourth photograph. The optic disc underwent a cup enlargement that was detected in color photographs at

the

third year.

15 had a progressive change in nerve fiber layer. Three of

these 8 were initially nerve fiber layer abnormal; hence,

10 of

the 15 were identified

by

a nerve fiber layer abnor

mality either initially or in follow-up.

ther

Parameters

A peripapillary crescent

was

detected in

27

of37 converter

eyes (73 ), in 20 of37 ocular hypertensive controls (54 ),

and in 49 of

88

visually healthy controls (56 ) (no dif

ferences significant at 0.05 level, chi-square analysis).

24

Among color photographs from all time points, a hem

orrhage near the disc margin occurred in 2 (5 ) converter

eyes, in

locular

hypertensive control

(3 ),

and in none

of the controls.

iscussion

We found that the predictive value of the cup-to-disc ratio

to identify glaucoma eyes that subsequently develop visual

field damage is less than ideal. J 9 J lOur data are similar


7/10

Quigley et al . Progression of Early Glaucoma Damage

Table 6. Nerve Fiber Layer Readings

in

Initial Photographs

onverters

Normal

Diffuse atrophy

Wedge defect

Unreadable

ontrols

Normal

Diffuse atrophy

Wedge defect

Unreadable

Superior

49

(18)

41

(15)

5 (2)

5 (2)

65 (24)

27

(10)

- 0 )

8 (3)

Inferior

57

(21)

35

(13)

3 (1)

5 (2)

68 (25)

24 (9)

- 0 )

8 (3)

Table 7. Type of Nerve Fiber Layer Change

in

Converters

Normal to mild diffuse

3

Normal to severe diffuse

3

Normal to mild wedge

3

Normal

to

severe wedge

3

Subtotal

12

Mild

to

severe diffuse

9

Mild to severe wedge

2

Subtotal

11

Total

23

to those of smaller studies that have evaluated cup-to-disc

ratio in persons followed prospectively to the point

of

field 10ss26-28 (Table 9). Whereas the mean cup-to-disc

ratio of converter eyes was larger than that

of

ocular hy

pertensive controls and visually healthy persons,

40

of

converter eyes would be missed by the best criterion for

separation, a vertical cup-to-disc ratio of at least 0.55.

More important, one fourth

of

ocular hypertensive eyes

that did not develop field loss over the 5-year follow-up

are labeled high risk by this criterion. We estimate that

this criterion, if applied to the ocular hypertensive pop

ulation before field loss, would identify

3

eyes that would

not develop field loss in 5 years for every 1 that would.

The nerve fiber layer evaluation was equal in sensitivity

and specificity to cup-to-disc ratio as a baseline criterion

for prediction

of future field loss. We previously reported

a

60

rate

of

nerve fiber layer abnormality 5 years before

field loss in this population.

21

The present data, in which

a proportion of these eyes were re-read independent of

the original evaluation, demonstrate the acceptable re

producibility of this evaluation.

Parameters other than cup-to-disc ratio were even less

predictive. Asymmetry between the cups in the two eyes

of 0.2 units identified only 1 of 10 converters and had an

estimated positive predictive power of only

6 .

This

means that

15

nonprogressive eyes are identified for every

future converter. Disc hemorrhages

9

were so infrequently

recognized in this population (as previously reported)30

that they provide little guidance. Vertically oval shape of

the cup also does not differentiate progressive and stable

eyes (Table 3). Peripapillary crescents were seen in half

of

normal and ocular hypertensive control eyes and in

three fourths of converters,

but

this difference was insig

nificant. Others have previously noted a somewhat greater

prevalence

of

crescents in glaucoma-damaged eyes than

in healthy eyesY Thus, other disc parameters are less

valuable than cup-to-disc ratio despite their reputation as

important clinical signs.

The disappointing performance of clinical disc ex

amination for prediction of future course has stimulated

research into image analysis methods to better define the

health of the optic nerve. Neural rim area

15

and topo

graphy of he nerve fiber layer surface

3

show promise, but

no prospective studies have been performed. Each requires

specialized equipment and substantial computation time.

Theoretically, a larger opening in the eyewall would

Table 8. Nerve Fiber Layer Correlated with

Humphrey

or

Goldmann

Field

Nerve Fiber Layer

Finding

by Hemifield

n Correct Both

rong

Normal

Unreadable

Goldmann

One

hemmeld abnormal 29 10

11

0 6 2

Both hemmelds abnormal

8

1

5 2 0

Humphrey

One hemmeld abnormal 20 8

5

1

5

1

Both hemmelds abnormal

10 0 8 1 1

Goldmann defect

met

criteria given

21

in

either one

or

both hemifields.

Humphrey

defect chosen for one hemifield abnormal was a glaucoma hemifield

test result (StatPac

II)

of outside normal limits (for one half of

the

field and,

in

addition,

in the

opposite half field th ere c ould

be

no more

than

2

points

in

total deviation plot exceeding normal values with P

An Evaluation of Optic Disc

Documents

Transcript of An Evaluation of Optic Disc