CLINICAL INVESTIGATION

Visual loss and perimetric sensitivity in eyes with retinitispigmentosa

Hiroyuki Iijima

Received: 14 January 2013 / Accepted: 27 May 2013 / Published online: 23 August 2013

� Japanese Ophthalmological Society 2013

Abstract

Purpose To investigate the various perimetric parameters

that best predict reduction of best-corrected visual acuity

(BCVA) to worse than 0.5 in the near future in eyes with

retinitis pigmentosa (RP).

Methods The most recent records obtained by Humphrey

Field Analyzer (HFA) central 10-2 perimetry were studied

for the right eyes of 123 patients (60 men and 63 women)

with typical RP. The correlation between various parame-

ters of perimetric sensitivity and BCVA was retrospec-

tively studied. The receiver operating characteristic (ROC)

curves were used to find the best parameter to discriminate

eyes with BCVA C0.5 from those with BCVA \0.5.

Results Spearman rank correlation coefficients with log-

MAR BCVA were the highest for the foveal threshold (FT)

and mean sensitivity of the test points within 1.4� of the fix-

ation point (MS1.4). The ROC curve analysis revealed that

the area under the curve was the largest for the MS1.4 among

all the perimetric parameters for discriminating eyes with

BCVA C0.5 from those with BCVA\0.5. The cutoff value of

30 dB showed 100 % specificity and 57 % sensitivity.

Conclusions The risk of vision decreasing below 0.5 in

the near future may be predicted when the mean sensitivity

within 1.4� of the fixation point in the HFA 10-2 reaches

30 dB in eyes with RP.

Keywords Retinitis pigmentosa � Perimetric

sensitivity � Visual loss � Humphrey perimetry �Receiver � Operating characteristic curve

Introduction

We have previously reported that deteriorating visual

function could be demonstrated using univariate linear

regression analysis of the mean deviation (MD) of Hum-

phrey Field Analyzer (HFA) 10-2 perimetry in eyes with

retinitis pigmentosa (RP) during follow-up for several

years with repeated HFA 10-2 perimetry [1].

We also recently conducted a cross-sectional study of

123 eyes with RP to identify the best parameter for mon-

itoring progression of RP among the MD, foveal threshold

(FT), and average sensitivity of the central 4 (CENT4) or

12 (CENT12) test points in HFA 10-2 perimetry, along

with the logarithm of the minimal angle of resolution

(logMAR) of the best-corrected visual acuity (BCVA). We

studied the correlation between these parameters and the

duration in years of the symptoms, defined as the number

of years since the patient became aware of night blindness

and/or narrowing of the visual field [2]. These investiga-

tions revealed that logMAR BCVA correlated less with the

duration in years of the symptoms than did the MD and

other measures of perimetric sensitivity, implying that

disease progression is better monitored by perimetric

parameters than by logMAR BCVA.

However, most patients with RP tend to be concerned

about their BCVA because those with a reduced peripheral

visual field are obliged to rely on the BCVA when working

at a desk, eating, watching television, or reading. Szlyk and

coworkers [3] reported that in patients with RP, BCVA is

more strongly correlated with patients&ratings of their dif-

ficulty in the performance of daily life than with the

residual visual field area, and that those with BCVA worse

than 0.5 have difficulty in performing daily tasks [4].

Therefore, in addition to observing the progression of

perimetric sensitivity, information on how many years they

H. Iijima (&)

Department of Ophthalmology, Faculty of Medicine,

University of Yamanashi, 1110 Shimokato, Chuo,

Yamanashi 409-3898, Japan

e-mail: hiijimar@yamanashi.ac.jp

123

Jpn J Ophthalmol (2013) 57:563–567

DOI 10.1007/s10384-013-0271-7

will be able to maintain a BCVA better than 0.5 is

important for these patients.

The aim of this study was to determine which of the

perimetric parameters could most accurately predict

decrease in BCVA to worse than 0.5. We reanalyzed the

data collected from a previous study [2] to study the rela-

tionship between logMAR BCVA and perimetric sensi-

tivity or its mean values in central fields of various sizes.

Participants and methods

The clinical data of 123 patients (60 men and 63 women)

with typical RP who participated in our previous study [2]

were retrospectively reviewed. The precise demographic

data have been reported previously [2].

RP was diagnosed on the basis of characteristic retinal

changes and a constricted visual field. In some cases,

markedly reduced or non-recordable electroretinograms

were used to confirm the diagnosis. Atypical cases,

including sectorial, central, unilateral, or secondary RP,

were excluded. Eyes with cystoid macular edema or an

epiretinal membrane were also excluded. Thirty-five eyes

with intraocular lenses without apparent posterior capsular

opacity were included.

The patients were followed up twice a year at the Uni-

versity of Yamanashi Hospital. In addition to measurement

of the BCVA and intraocular pressure and slit lamp and

fundus examinations, we routinely conducted Humphrey

Field Analyzer (HFA) 10-2 or 30-2 perimetry using the

Fastpac protocol (HFA; Humphrey Instruments, San

Leandro, CA, USA) to monitor progression of RP. The

HFA perimetry results were primarily used to provide the

patients with information on the progression of the disease.

Of the many sets of HFA 10-2 results for each patient, the

most recent result with sufficient reliability in the right eye

was used for the present study.

In addition to the MD and FT, we also investigated the

mean sensitivity of 9 different central field areas. Sixty-eight

test points in the HFA 10-2 display were grouped into 9

equidistant groups from the fixation point, as indicated in

Fig. 1. The numbers 1.4, 3.2, 4.2, 5.1, 5.8, 7.1, 7.6, 8.6, and 9.1

represent the degrees from the fixation point. The mean sen-

sitivity was then obtained for 9 central field areas within cir-

cles with different radii (Table 1). The mean sensitivity in the

present study is theoretically proportional to the ‘‘total score

point’’ reported by Berson and coworkers [5, 6], although the

HFA field size was different: 10� in the present study, but 30 or

60� in the reports by Berson and coworkers. Although MS9.1

is the mean sensitivity at all 68 points in the HFA 10-2 pro-

gram, it does not precisely correspond to the HFA 10-2 MD

because the latter is not a simple mean but a weighted mean of

the total deviation at the same 68 test points.

The BCVA examined on the same day as the HFA

perimetry testing using a standard Japanese decimal visual

acuity chart was converted into the logMAR. The corre-

lation coefficients between the logMAR BCVA and the 11

sensitivity measures, including the FT, MD, and 9 mean

sensitivities of the different central field areas, were

determined.

This study, using clinical data, including visual acuity

testing and perimetry results, was approved by the Uni-

versity of Yamanashi Hospital ethics committee and con-

formed with the Declaration of Helsinki.

The statistical analyses were conducted using IBM

SPSS Statistics software (version 19.0; SPSS, New York,

NY, USA). StatFlex (version 6; Artec Company, Osaka,

Fig. 1 Distance from the fixation point in the HFA central 10-2

display. Figures designate the distance (degrees) from the fixation

point (closed circle) for 68 test points of the HFA 10-2 display

Table 1 Mean sensitivity (MS) of 9 different central field areas

Mean sensitivity of

each field area

Degrees of test points from

the fixation point

Number of

points

MS1.4 1.4 4

MS3.2 B3.2 12

MS4.2 B4.2 16

MS5.1 B5.1 24

MS5.8 B5.8 32

MS7.1 B7.1 44

MS7.6 B7.6 52

MS8.6 B8.6 60

MS9.1 B9.1 68

564 H. Iijima

123

Japan) was used to draw receiver operating characteristic

(ROC) curves to find the best parameters to discriminate

eyes with BCVA C0.5 from those with BCVA \0.5.

Results

Scattergrams showing the relationships between the FT,

MS1.4, and MD and the logMAR BCVA are shown in

Fig. 2. The FT correlated more highly with the logMAR

BCVA than did the MD throughout the whole range of the

logMAR BCVA values. The correlation between MS1.4

and logMAR BCVA appeared high within a narrow range

of low logMAR BCVA values (i.e., good BCVA), whereas

it appeared poor within the range of logMAR BCVA of 0.3

or higher.

The Spearman rank correlation coefficients with the

logMAR BCVA in all 123 eyes are presented as gray bars

in Fig. 3 for the FT, 9 mean sensitivities of the different

central field areas, and MD. The correlation coefficient

with the logMAR BCVA was the highest for the FT

(q = -0.794, P \ 0.001) and the lowest for the MD

(q = -0.533, P \ 0.001). The correlation coefficients

between the logMAR BCVA and the 9 mean sensitivities

of the different central field areas were intermediate in a

manner dependent on the narrowness of the field.

To predict a decrease in BCVA to worse than 0.5 in the

near future, the correlation between various sensitivity

measures and logMAR BCVA should be studied in eyes

without extremely poor BCVA. Thus, the results in a

subset of participant eyes without poor BCVA, consisting

of 100 eyes with a logMAR BCVA of 0.6 or lower (0.25 or

better BCVA), are also presented as black bars in Fig. 3.

As such, the order of the higher Spearman rank correlation

coefficient with the logMAR BCVA was altered, with

MS1.4 the highest (q = -0.719, P \ 0.001) instead of the

FT (q = -0.671, P \ 0.001).

The best parameters for predicting future BCVA loss to

worse than 0.5 were investigated using ROC curves. The

ROC curves of the 3 candidate parameters including MD,

FT, and MS1.4 to distinguish eyes with BCVA equal to or

better than 0.5 from those with BCVA worse than 0.5 are

demonstrated in Fig. 4. The diagnostic performance could

be judged by the closeness of the ROC curve to the upper

left corner of the graph or the area under the curve (AUC).

Table 2 shows the AUCs of the FT, the mean sensitivities

including MS1.4 through MS9.1, and the MD. The AUC

for MS1.4 was the largest, although the AUCs for MS1.4

and FT did not differ significantly (P = 0.017).

We examined the cutoff values to discriminate eyes with

BCVA equal to or better than 0.5 from those with BCVA

worse than 0.5 under the requirement of 100 % specificity

Fig. 2 Scattergrams of the FT (a), MS1.4 (b), and MD (c) vs

logMAR BCVA. The Spearman rank correlation coefficients with

logMAR BCVA were -0.794, -0.788, and -0.533 for the FT,

MS1.4, and MD, respectively (P \ 0.001). The solid lines indicate

the linear regression lines. The vertical dashed lines indicate a

logMAR BCVA of 0.3, which is equivalent to a BCVA of 0.5. The

horizontal dashed lines indicate 34, 30, and -8 dB for the FT, MS1.4,

and MD, respectively

Fig. 3 Spearman rank correlation coefficient between logMAR

BCVA and each of the perimetric parameters. The bars represent

the Spearman rank correlation coefficient with the logMAR BCVA

for the FT, the 9 mean sensitivities of the different central field areas,

and the MD in all 123 participants (gray bars) and in the subset of 100

eyes with logMAR BCVA B0.6 (black bars)

Perimetric sensitivity in RP 565

123

and found 34, 30, and -8 dB cutoff values for FT, MS1.4,

and MD, respectively (shown as horizontal dashed lines in

Fig. 2). The sensitivities at these cutoff values were 0.152,

0.570, and 0.303 for the FT, MS1.4, and MD, respectively.

Discussion

The HFA 10-2 mean deviation has been used to assess the

progression of RP and to investigate the effectiveness of

oral nilvadipine in retarding its progression [7]. However,

the total point score with the HFA 30-2, which is the sum

of all sensitivity values and is almost equivalent to the HFA

30-2 MD, did not demonstrate effectiveness of docosa-

hexaenoic acid or lutein supplement in slowing the pro-

gression of the RP in patients who were also receiving

vitamin A [5, 6]. This failure to show the effectiveness of

the supplements to slow RP progression may be due to the

relatively low sensitivity of the HFA 30-2 MD in moni-

toring RP progression. Averaging or summing all of the

sensitivities at the 74 test points in the HFA 30-2 test,

which usually includes many test points of null decibels,

especially outside the 10� central field, may dilute mean-

ingful sensitivity values at test points within the 10� central

field and weaken its power to detect progression of RP. We

consider that the HFA 10-2 MD is superior to the HFA

30-2 MD in detecting the progression of RP because

absolute scotoma is frequently observed outside the 10�central field in many patients with RP in clinical studies

[2]. HFA 10-2 perimetry has also been reported to be

preferable to 24-2 or 30-2 perimetry in eyes with advanced

glaucoma [8].

With regard to the progressive decrease in perimetric

sensitivity and visual loss below 0.5, we previously

reported that eyes with HFA 10-2 MD of -15 dB or lower

had an increased likelihood of exhibiting BCVA values

below 0.5 [9]. However, in a more recent study, we dem-

onstrated that FT and the mean sensitivity at 4 or 12 central

test points, designated as CENT4 and CENT12, respec-

tively, correlated more highly with the logMAR BCVA

than did the HFA 10-2 MD [2]. The CENT4 and CENT12

in that previous study were compatible with the MS1.4 and

MS3.2 used in the present study. Indeed, as shown in the

scattergram of the MD and logMAR BCVA in Fig. 2, some

eyes with MD [-15 dB showed logMAR BCVA values

[0.3 (i.e., BCVA lower than 0.5).

In the current study, we studied the Spearman rank

correlation coefficient with the logMAR BCVA for various

parameters of visual sensitivities and found that the highest

Spearman rank correlation coefficient with logMAR

BCVA was MS1.4 in 100 eyes without extremely poor

BCVA. Further analysis using the AUC of the ROC curve

showed that the power to detect BCVA equal to or better

than 0.5 was the strongest for MS1.4 showing the largest

AUC. If 30 dB of MS1.4 is chosen as the cutoff value to

discriminate eyes with BCVA equal to or better than 0.5

from those with BCVA worse than 0.5, the specificity is

100 % and the sensitivity is sufficiently high (57 %), to

allow it to be effectively used to advise patients with RP in

terms of predicting visual loss below 0.5 in the near future.

The usefulness of MS1.4 with a threshold of 30 dB may be

confirmed by a longitudinal follow-up study to investigate

the relationship between progressive decrease in MS1.4

and BCVA changes in individual patients with RP.

Fig. 4 Receiver operating characteristic curves for detection of a

BCVA equal to or better than 0.5. Curves for the FT (solid black line),

MS1.4 (dashed black line), and MD (solid gray line) are drawn for all

the participant eyes

Table 2 Diagnostic performance of perimetric sensitivity

Perimetric sensitivity AUC ± SE

FT 0.909 ± 0.028

MS1.4 0.911 ± 0.025

MS3.2 0.883 ± 0.029

MS4.2 0.873 ± 0.031

MS5.1 0.856 ± 0.033

MS5.8 0.841 ± 0.035

MS7.1 0.825 ± 0.037

MS7.6 0.814 ± 0.039

MS8.6 0.809 ± 0.039

MS9.1 0.806 ± 0.040

MD 0.792 ± 0.040

AUC area under the curve, SE standard error, FT foveal threshold, MS

mean sensitivity, MD mean deviation

566 H. Iijima

123

In conclusion, MS1.4, which is the mean sensitivity of 4

central test points in the HFA 10-2 program, is a good

indicator for monitoring the progression of advanced RP

and is also useful in predicting the risk of visual loss in the

near future if it reaches 30 dB.

Acknowledgments This study was supported in part by a JSPS

KAKENHI Grant (no. 22591937) (Grant-in-aid for scientific

Research [C]) from the Japan Society for the Promotion of Science.

The author has no financial conflicts of interest to declare. The author

thanks the members of the Retina Clinic of the University of

Yamanashi Hospital, including Kyoko Kohno, Daisuke Yuzurihara,

Toshihito Ariizumi, Tadasuke Hatori, Nami Chiba, Yoichi Sakurada,

and Naohiko Tanabe, for their contributions in examining the par-

ticipating patients. The author also thanks Naoko Matsui, Miho

Sakurabayashi, Keiko Hanada, Takako Tanaka, and Yuki Suzuki for

their assistance in conducting HFA perimetry.

References

1. Hirakawa H, Iijima H, Gohdo T, Imai M, Tsukahara S. Progression of

defects in the central 10� visual field of patients with retinitis

pigmentosa and choroideremia. Am J Ophthalmol. 1999;127:436–42.

2. Iijima H. Correlation between visual sensitivity loss and years

affected for eyes with retinitis pigmentosa. Jpn J Ophthalmol.

2012;56:224–9.

3. Szlyk JP, Fishman GA, Alexander KR, Revelins BI, Derlacki DJ,

Anderson RJ. Relationship between difficulty in performing daily

activities and clinical measures of visual function in patients with

retinitis pigmentosa. Arch Ophthalmol. 1997;115:53–9.

4. Szlyk JP, Seiple W, Fishman GA, Alexander KR, Grover S,

Mahler CL. Perceived and actual performance of daily tasks:

relationship to visual function tests in individuals with retinitis

pigmentosa. Ophthalmology. 2001;108:65–75.

5. Berson EL, Rosner B, Sandberg MA, Weigel-DiFranco C, Moser

A, Brockhurst RJ, et al. Clinical trial of docosahexaenoic acid in

patients with retinitis pigmentosa receiving vitamin A treatment.

Arch Ophthalmol. 2004;122:1297–305.

6. Berson EL, Rosner B, Sandberg MA, Weigel-DiFranco C,

Brockhurst RJ, Hayes KC, et al. Clinical trial of lutein in patients

with retinitis pigmentosa receiving vitamin A. Arch Ophthalmol.

2010;128:403–11.

7. Nakazawa M, Ohguro H, Takeuchi K, Miyagawa Y, Ito T, Metoki T.

Effect of nilvadipine on central visual field in retinitis pigmentosa: a

30-month clinical trial. Ophthalmologica. 2011;225:120–6.

8. Much JW, Liu C, Piltz-Seymour JR. Long-term survival of central

visual field in end-stage glaucoma. Ophthalmology. 2008;115:1162–6.

9. Abe K, Iijima H, Hirakawa H, Tsukahara Y, Toda Y. Visual acuity

and 10� automated static perimetry in eyes with retinitis pigmen-

tosa. Jpn J Ophthalmol. 2002;46:581–5.

Perimetric sensitivity in RP 567

123