5 (Blue) Cone Pathway Vulnerability in Retinitis Pigmentosa, Diabetes and Glaucoma

Vivienne C. Greenstein,* Donald C. Hood,t Robert Rirch4 David Sreinberger4 and Ronald E. Carr*

A variety of retinal diseases lead to a decrease in the sensitivity of the S (blue) cone pathways. To determine the possible sites and mechanisms of this loss we compared the sensitivities of an S (blue/pi-1) and an M (green/pi-4) cone pathway in patients with retinal diseases that differ as to their primary locus of sensitivity loss. The sensitivities of an S and an M cone pathway were assessed in patients with retinitis pigmentosa, insulin-dependent diabetes mellitus and open-angle glaucoma using Stiles two-color increment threshold technique. A greater loss in sensitivity of an S than an M cone pathway was found for all three disease groups; however, the diabetic patients showed a more selective loss. The results suggest that multiple sites are involved and that the combined effects of metabolic abnormalities and hypoxia contribute to the selective loss. Invest Ophthalmol Vis Sci 30:1732- 1737,1989

A selective loss in sensitivity to short-wavelength The three diseases do not necessarily produce the light has been reported in diseases as diverse as retini- same degree of relative S cone pathway loss and a tis pigmentosa (RP), diabetes mellitus and glau- comparative study may provide some understanding coma.'"7 This selective loss has been attributed to of the factors that contribute to the apparent vulnera- vulnerability of the S "blue" cone photoreceptors to bility or selective sensitivity loss of the S cone path- disease. However, the presence of this loss in sensitiv- way. We compared the pattern of S and M cone ity in such diverse diseases suggests that it may in part pathway sensitivity losses in patients with these three be attributed to involvement of post-receptoral sites. diseases using Stiles two-color increment threshold Until the affected anatomic locus is identified, the technique. The sensitivities of the cone pathways are loss is better described as a selective sensitivity loss of often compared using techniques that isolate each the S cone pathway. cone pathway with a different adapting field. The as- In retinitis pigmentosa, increment threshold data sumption underlying these techniques is that sensitiv- show a greater loss in sensitivity for an S cone path- ity losses in retinal disease are independent of the way (pi-1) than for an M "green" cone pathway level of adaptation. Greenstein et al,8 however, have (pi-4).2-4 A similar selective loss has been reported for shown that sensitivity losses can depend on the level patients with insulin-dependent diabetes mellitus of adaptation used. Consequently in this study we (IDDM) who have minimal or no apparent retinop- obtain measures at several levels of adaptation. A athy,3-5 and for patients with glaucoma.67 These three preliminary report of some of the data for RP and disease groups differ as to their primary retinal locus IDDM patients has been published elsewhere.8 of sensitivity loss. RP is a disease primarily of the photoreceptors, glaucoma a disease of the inner ret- Materials and Methods ina, the ganglion cell layer, whereas diabetic retinopathy affects both mid and inner retinal layers, and Subjects perhaps even the outer retinal layer. Twenty-three patients with RP, 14 with IDDM and 16 (18 eyes) with open-angle glaucoma (OAG; either From the *Department of Ophthalmology, New York University primary open-angle glaucoma or pigmentary glau- Medical Center, New York, the fDepartment of Psychology, Co- lumbia University, New York and the JNew York Eye and Ear coma) participated in the study. All patients had Infirmary, New York, New York. Snellen acuity of 20/40- or better. Of the 23 RP Supported by NIH Grant EY-02115 and a grant from RP Foun- patients, ten were classified as autosomal recessive, dation Fighting Blindness. one as autosomal dominant, one as X-linked, and 11 Submitted for publication: January 11, 1989; accepted February as RP simplex. None of the RP patients showed evi- 7, 1989. Reprint requests: Vivienne C. Greenstein, Department of Oph- dence of cystoid macular edema or significant lens thalmology, New York University Medical Center, 550 First Ave- opacities. Using full-field nonsummated single-flash nue, New York, NY 10016. ERGs, all had nonrecordable or markedly reduced

1732

Downloaded from iovs.arvojournals.org on 09/25/2021 No. 8 S (DLUE) CONE PATHWAY VULNERABILITY IN RETINAL DISEASE / Greensrein er ol 1733

scotopic and photopic ERGs. Visual fields were mea- Table 1. Clinical findings sured on a Goldmann perimeter using a III/4 "white" test object; and the area of the remaining visual field RP n = 23 Age range 14-55 yrs was estimated; none of the patients studied had a Total area of remaining visual field visual field <10° (see Table 1 for details). 10°-20° 15 20°-30° 4 Of the 14 patients with IDDM, clinical examina- 30°-40° 4 tion and fluorescein angiography revealed evidence IDDM n = 14 Age range of early background retinopathy in 13 patients (see 24-60 yrs Table 1 for level of retinopathy using a modified Air- Level 1 1 9 Level 2 4 lie House classification ). Three of the patients had Level 3 5* macular edema (two at retinopathy level 3, and one Level 4 4* at retinopathy level 4). OAG n = 18 Age range 29-65 yr The patients with OAG had varying degrees of field Field loss loss and cupping. The patients had mild to moderate No appreciable loss 2 glaucomatous damage. Three patients had scotomas Defect 0°-5° from fixation 2 Defect 5°-10° from fixation 1 extending to within 10° of fixation on repeated test- Defect 10°-30° from fixation 3 ing using program 32 on the Octopus 201 R comput- Defects 5°-10°, and 10°-30° from fixation 2 erized perimeter (Interzeag, Schleiren, Switzerland) Defects 0°-5°, 5°-10°, and 10°-30° from 8 fixation (see Table 1). Intraocular pressure was controlled clinically in all patients at the time of examination. Pupil diameters ranged from 2 to 5 mm. Patients with at least 2 min before the test light was presented. For pupil diameters less than 2 mm were excluded from an ascending run, test intensity was increased from the study. below threshold to the level at which two consecutive Eleven subjects ranging in age from 20-55 years presentations were visible, and for a descending run it with no known abnormality of the visual system was decreased to the level at which two consecutive comprised the control group. They were divided into presentations were no longer visible. Test spectral two groups by age (20-40 years and 40-55 years). sensitivities were obtained in selected patients to con- Informed consent was obtained from all subjects. firm the mechanism mediating detection. The use of a 480 nm test light allowed for a measure of the rela- Apparatus tive M and S cone pathway sensitivity loss unaffected by differences in preretinal absorption. For each sub- Light stimulation was provided by a two-channel 10 ject the contribution of macular pigment density to Maxwellian view system previously described. The threshold measures was estimated by comparing the final lens formed 1.8 X 2.0 mm rectangular images of value of M cone pathway thresholds obtained with the source. Monochromatic light was provided by in- the 480 nm test light at an adapting field intensity of terference filters with half bandwidths of approxi- 0.96 log Td to those obtained with a 540 nm test light. mately 6 nm. Retinal illumination was calibrated If the decreased M cone sensitivity measured with a with an EG and G model 550 photometer (Salem, 480 nm test light was due to the presence of increased MA) and calculated using the method described by macular pigment density, then this decrease should Westheimer." be negligible for data obtained with a 540 nm test light. The absorption at 540 nm is minimal even in Procedure subjects with high density macular pigment.13 All A two-color increment threshold procedure was thresholds were adjusted to allow for the contribution used to measure M (green) and S (blue) cone pathway of macular pigment. sensitivity. Foveal increment thresholds were obtained for a 480 nm test light (1°, 200 msec) super- Results imposed on 14°, 600 nm steady adapting fields. The The mean foveal increment threshold data for 11 conditions were selected to measure Stiles' pi-4 (an M normals are shown in Figure 1. The filled triangles cone pathway) and pi-1 (an S cone pathway) mecha- 12 represent mean thresholds for six normals in the nisms. Following 10 min of dark adaptation, com- 20-40 year age group and the open circles mean plete increment threshold curves were obtained in thresholds for five normals in the 40-55 year group. approximately 30 min on each patient. Threshold Error bars represent ± 1 SD of the normal means. was defined as the mean of three to five repetitions of Because of the similarity of the thresholds for both an ascending and descending modified method of groups of normals, the patients' data were compared limits. The subject adapted to each adapting field for to the mean thresholds for all the normals shown in

Downloaded from iovs.arvojournals.org on 09/25/2021 1734 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / August 1989 Vol. 00

IQ: ().96 logTd Io- 2.91 logTd 2 s 6 2- *"x 5 >» >- 1- A A /\ f /

in - • Z.M Cone 4 pigment 1 >-SCone | pigment V) SENSITIVI

-1 - -OG i 4 1 1 i 3 D) 400 500 600 400 500 600 O Fig. 3. Test spectral sensitivities obtained at two levels of adapta- -1 tion for one normal subject. The dashed curves are estimates of the Log Adapting Intensity (td) S and M cone action spectra based on the Smith and Pokorny cone (600 nm) fundamentals.20 Fig. 1. Mean foveal increment threshold data for two groups of normals for a 480 nm test light on a series of 600 nm adapting patient, the open symbols are the mean thresholds for fields. The filled triangles represent mean threshold for six normals normals from Figure 2. For each patient the normal in the 20-40 year age group, and the open circles are data for five normals in the 40-55 year group. Error bars represent ±1 SD. curve has been moved vertically to fit the data (dashed curve). The RP patient (Figure 4A) shows a decrease in sensitivity of the M cone pathway of ap- Figure 2. The two curves are the best fit of a function proximately 0.5 log unit at 0.96 log Td and a decrease commonly used to describe threshold vs. intensity 14 in sensitivity of the S cone pathway of 1.0 log unit. (tvi) data. It is well documented that the lower The IDDM patient (Fig. 4B), however, shows little if branch represent detection mediated by the M cone any change in sensitivity of the M cone pathway and pathway, and the upper branch detection mediated 14 a marked loss in sensitivity of the S cone pathway. by the S cone pathway. Test spectral sensitivities Yet another pattern of loss is seen for the OAG pa- obtained at 0.96 log Td and 2.91 log Td are shown in tient (Fig. 4C); the decrease in sensitivity is approxi- Figure 3 (similar data were obtained for other nor- mately equal for both the M and S cone pathways. mals). The spectral sensitivity data confirm that de- Foveal increment threshold curves were obtained tection of the 480 nm test light is mediated by ab- on all patients. The sensitivity loss of the M cone sorption in the M cones at low levels of adapting pathway and the S cone pathway compared to the intensity and by absorption in the S cones at the normal was calculated. The log decrease in sensitivity higher levels (see dashed vertical lines in Fig. 3). at 0.96 log Td (a measure of the M cone pathway) was Representative foveal increment threshold data for compared to the decrease at 3.87 log Td (a measure of patients with RP, IDDM and OAG are shown in Fig- the S cone pathway). These adaptation levels were ure 4. The filled symbols are data for an individual chosen to ensure that we could measure sizable losses in sensitivity yet stay within the M and S cone pathways (see Fig. 4A). They also resulted in approxi- 2.5- mately equal states of adaptation in both pathways.

V) Different states of adaptation could give rise to mis- leading conclusions about disease-related sensitivity E E losses.8 Eighteen of the patients with RP, ten of the patients with IDDM and 15 of the 18 eyes of patients 2g 0.5- with OAG showed either M or S cone pathway losses greater than 0.2 log unit. A comparison of these sensi- -0.5- tivity losses for the three groups of patients is shown in Figure 5. The diagonal lines represent the locus of O) -1.5 equal M and S cone pathway losses. Each data point o1 - -10 12 3 4 5 represents an individual patient's loss in sensitivity Log Adapting Intensity (td) compared to the normal. Most of the data fall to the (600 nm) right of the diagonal, evidence for a selective loss in Fig. 2. Mean foveal increment threshold data for 11 normals. sensitivity of the S cone pathway. Thirteen out of 18 The two branched curve is the best fito f the equation log T = log T o RP patients (72%), 12 out of 15 OAG patients (80%) + log ((Ao + A)/Ao), where T is the test threshold, To is the dark adapted test threshold, Ao is a constant, and A is the adapting and all the diabetics show a greater loss in sensitivity intensity of the 600 nm background. of the S than the M cone pathway. Only five patients

Downloaded from iovs.arvojournals.org on 09/25/2021 No. 8 S (BLUE) CONE PATHWAY VULNERABILITY IN RETINAL DISEASE / Greensrein er ol 1735

B. A RP IDDM 3 - Normals A A Normal y** [td) 2 - ..A'*A >> 2-

Vi c c © ~- 1 - E 1 - c si ^- o 18 oo 0 - 0 - V) f->^ 0) T . .c -1 - ~F -^ -1 - -4- en cn o o • 1 • 1 I 1 I 1 • 1 1 • 1 -2- - 1 0 12 3 4 5 0 12 3 4 Log Adapting Intensity (td) Log Adapting Intensity (td) (600 nm) (600 nm)

c. 3 - • OAG S" *-> A Normal y >. 2- '55 c Fig. 4. (A) Increment threshold data for an RP patient compared 1 - to normals. The solid curve is the increment threshold curve for Inte nm) normals from Figure 2; it has been moved vertically to fit the o A,..-*' >C patients' data (dashed curve). (B) Increment threshold data for an O CO 0 - ""A ^x IDDM patient compared to normals. (C) Increment threshold data CD for an OAG patient compared to normals. —*• . C -1 -

D) O .9 - - 1 0 1 2 3 4 Log Adapting Intensity (td) (600 nm)

show a greater loss in sensitivity of the M cone path- cone pathway than the RP or OAG patients; all but way (three with RP and two with OAG). one of the ten diabetics showed either no loss or a The loss in S cone pathway sensitivity appears to be slight M cone pathway loss (<0.3 log unit). The pat- more selective for the diabetics than for the RP or tern of S and M cone losses for RP and OAG patients OAG patients. S cone pathway losses for the majority was similar. of the diabetics are accompanied by either slight If S cone pathway vulnerability could be attributed (<0.3 log unit) or no M cone pathway losses. Only to the receptors then one might expect that RP rather three of these patients were observed to have macular than diabetes or OAG would result in a more selec- edema. tive loss in S cone pathway sensitivity. RP comprises a group of heredodegenerative retinal disorders, pro- Discussion gressive in nature, that primarily affect the receptors and the RPE. The RP patients do not show a more The S cone pathways are said to be more vulnera- selective loss. Similarly, if the locus for vulnerability ble to retinal disease than the other cone pathways, were at or near to the ganglion cell layer OAG pa- and there is ample evidence to support this in the tients should show the more selective loss. This is not literature.15 What is the locus and mechanism of this the case. In fact the RP and OAG patients exhibit a vulnerability? Can the sensitivity loss be attributed to similar pattern. The finding that all three diseases some mechanism at the level of the receptors or is the result in decreased S cone pathway sensitivity implies locus post-receptoral? The greater loss in sensitivity that multiple sites are probably involved; it is unlikely of an S cone pathway, found for all three groups, is in that the decreased sensitivity can be attributed solely agreement with previous studies. Diabetics, however, to S cone photoreceptors. Further, the magnitude of showed a more selective loss in sensitivity of the S the effect found in diabetic patients with mild back-

Downloaded from iovs.arvojournals.org on 09/25/2021 1736 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / Augusr 1989 Vol. 30

A. RP

1.4

1.2 Loss >» 1.0

0.8 isitivi 0) c 0.6 (/) O) (A) 0) 0.4 c

Cone o O S 0.2

Log 0.0 o

-0.2 •—I i 1 1 1—- 1 r-ll 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Log S Cone Sensitivity Loss Log S Cone Sensitivity Loss

Fig. 5. (A) A comparison of the log sensitivity loss of the S cone pathway to that of the M cone pathway for 18 RP patients. The diagonal line represents the locus of equal losses of the M and S cone pathways. The dashed vertical and horizontal lines represent the normal range. (B) The comparison for ten IDDM patients. (C) The comparison for 15 eyes of OAG patients.

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Log S Cone Sensitivity Loss

ground retinopathy suggests that the mechanisms of old would be more affected than that of the L or M sensitivity loss are more relevant than the sites of the cone pathways.17 Part of the S cone losses seen in the disease-related sensitivity loss. It is possible that the RP and OAG patients may also be due to hypoxia. combined effects of metabolic abnormalities and There are reports of the development of neovascular- hypoxia associated with diabetes contribute to the ization in RP patients, possibly due to relative hyp- selective loss. There are reports of selective sensitivity oxia,18 and abnormal autoregulation of macular reti- losses in diabetic patients with no retinopathy, that is, nal blood flow may affect the susceptibility of the eye before clinically observable retinal vascular changes to glaucomatous damage.19 occur,3 and hypoxia is reported to result in tritan-like defects in normals.16 One possible explanation for Key words: S cone pathway, retinitis pigmentosa, diabetes, this selective vulnerability of an S cone pathway to glaucoma, M cone pathway hypoxia comes from a psychophysical study showing that the response range of the S cone pathways is References more limited. Hypoxia could have its effect by de- 1. Marre E and Marre M: The influence of the three color vision- creasing the maximum response of all the cone path- mechanisms on the spectral sensitivity of the fovea. Mod Probl ways by an equal amount; however, as the response Ophthalmol 11:219, 1972. range of S cone pathways is more limited, the thresh- 2. Sandberg MA and Berson EL: Blue and green cone mecha-

Downloaded from iovs.arvojournals.org on 09/25/2021 No. 8 S (DLUE) CONE PATHWAY VULNERABILITY IN RETINAL DISEASE / Greensrein er ol 1737

nisms in retinitis pigmentosa. Invest Ophthalmol Vis Sci ence the detection of small, brief lights. Vision Res 22:89, 16:149, 1977. 1982. 3. Zwas F, Weiss H, and McKJnnon P: Spectral sensitivity mea- 11. Westheimer G: The Maxwellian view. Vision Res 6:669, 1966. surements in early diabetic retinopathy. Ophthalmic Res 12. Stiles WS: Separation of the "blue" and "green" mechanisms 12:87, 1980. of foveal vision by measurements of increment thresholds. 4. Young RSL: Early-stage abnormality of foveal pi mechanisms Proc Rov Soc B 133:96, 1946. in a patient with retinitis pigmentosa. J Opt Soc Am 72:1021, 13. Bone RA and Sparrock JMB: Comparison of macular pigment 1982. densities in human eyes. Vision Res 11:1057, 1971. 5. Adams AJ, Zisman F, Ai E, and Bresnick G: Macular edema 14. Wyszecki G and Stiles WS: Color Science. New York, John reduces B cone sensitivity in diabetics. Applied Optics 26:1455, Wiley & Sons, 1967. 1987. 15. Pokorny J, Smith VC, Verriest G, and Pinckers A (Eds.): Con- 6. Drum B, Armaly MF, and Huppert WE: Sources of short genital and Acquired Color Vision Defects. New York, Grune wavelength sensitivity loss in glaucoma. Doc Ophthalmol Proc &Stratton, 1979. Ser46:413, 1987. 16. Smith VC, Ernest JT, and Pokorny J: Effect of hypoxia on FM 7. Adams AJ, Rodic R, Husted R, and Stamper R: Spectral sensi- 100-hue test performance. Mod Probl Ophthalmol 17:248, tivity and color discrimination changes in glaucoma and glau- 1976. coma-suspect patients. Invest Ophthalmol Vis Sci 23:516, 17. Hood DC, Benimoff NI, and Greenstein VC: The response 1982. range of the blue-cone pathways: A source of vulnerability to 8. Greenstein VC, Hood DC, and Carr RE: A comparison of disease. Invest Ophthalmol Vis Sci 25:864, 1984. "blue" S-cone sensitivity loss in patients with diabetes and RP. 18. Uliss AE, Gregor ZJ, and Bird AC: Retinitis pigmentosa and In Colour Vision Deficiences IX, Drum B and Verriest G, retinal neovascularization. Ophthalmology 93:1599, 1986. editors. Dordrecht, KJuwer Academic Publishers, 1989, pp. 19. Grunwald JE, Riva CE, Stone RA, Keates EU, and Petrig BL: 233-241. Retinal autoregulation in open-angle glaucoma. Ophthalmol- 9. Klein BEK, Davis MD, Segal P, Long JA, Harris A, Haug GA, ogy 91:1690, 1984. Magli YL, and Syrjala S: Diabetic retinopathy assessment of 20. Smith VC, Pokorny J, and Starr SJ: Variability of color mix- severity and progression. Ophthalmology 91:10, 1984. ture data: 1. Interobserver variability in the unit co-ordinates. 10. Finkelstein M and Hood DC: Opponent-color cells can influ- Vision Res 16:1087, 1976.

Downloaded from iovs.arvojournals.org on 09/25/2021