Eye (1 991) 5, 362-368

Electroretinography as a Prognostic Indicator or Neovascularisationin CRVO

A. J. MORRELL*, D. A. THOMPSON,t J. M. GIBSON*, E. E. KRITZINGER*, N. DRASDOt Birmingham

Summary A prospective study was carried out to compare the efficacy of , fundus and clinical examination in identifying those at risk of rubeotic following central retinal vein occlusion (CRVO). Our prelimi­ nary observations are described. The findingssuggest a complementary role for elec­ troretinography in the management of CRVO. Particularly significant were interocular differences in 30Hz flicker latency, the ability to elicit pattern ERGs and the ratio of scotopic and photopic a: b w�ve amplitudes. Of the clinical measures the depth of the relative afferent pupil defect was a sensitive indicator of rubeosis. Fac­ tors oflesser statistical prognostic value included poor visual acuity and the extent of deep retinal haemorrhages. Fundus fluorescein angiography in this study had lim­ ited value in predicting those patients at risk of rubeosis.

The development of neovascularisation and develop complications.7 This largely reflects secondary glaucoma is a significantcomplica­ the limitations of fundus fluorescein angio­ tion of central retinal vein occlusion (CRVO) graphy: assessment of the degree of retinal occurring in 17-33% of patients.1,2,3 It is non-perfusion on the angiogram can be tech­ essential that those patients who are at risk of nically difficult due to factors such as small developing new vessels are identified at an pupil size, media opacities and the presence of early stage as prophylactic laser pametinal extensive deep retinal haemorrhages obscur­ photocoagulation has been shown to be effec­ ing the retinal circulation. Even with good tive in eliminating the neovascular quality angiograms inconsistencies of inter­ response.4,5 pretation occur amongst experienced It has been suggested that the stimulus for observers.8 new vessel growth is a vasoproliferative factor New vessel growth in venous occlusion is released by ischaemic retina.6 At present considered to be a response to hypoxia rather clinical and fluorescein angiographic features than retinal cell death.1,2,6 Fluorescein angio­ are used to determine the degree of retinal graphy identifies under-perfused areas of ret­ ischaemia present and to predict those ina, but these may be either hypoxic or which will develop neovascularisation. How­ infarcted retina. It is possible that the elec­ ever, these methods lack specificity,and result troretinogram (ERG), which can distinguish in the prophylactic treatment of a much larger decreased retinal sensitivity from non-func­ number of patients than would be expected to tioning retina, may be a better means of pre-

From: *Birmingham and Midland Hospital, Birmingham UK, tVision Sciences, Aston University, Birmingham, UK.

Correspondence to: A. J. Morrell, Birmingham and Midland Eye Hospital, Church Street, Birmingham B3 2NS. ELECTRORETINOGRAPHY AS A PROGNOSTIC INDICATOR OR NEOVASCULARISATIONIN CRVO 363 dicting neo-vascularisation. The a- and b­ cular disorders were excluded. The mean age wave components of the ERG respectively of the group was 61 years, range 29-87 years. reflect the functional integrity of the recepto­ All patients were seen in the Medical Oph­ ral and inner nuclear layers, which in turn thalmology Clinic within three weeks of pres­ depend upon differing blood supplies. Some entation for clinical assessment and fundus earlier studies of the ERG in CRVO have fluorescein angiography. Electroetinography demonstrated that patients with non-perfused was carried out with 7-12 days of (usually CRVO, had reduced b/a wave amplitude prior to) the fundus fluorescein angiogram. ratios.9,10 It is possible this ratio may be a sen­ Informed consent was obtained from all sitive index of perfusion from the retinal cir­ patients entered into the study: (one patient culation, because of the differential locus of refused consent for fluorescein angiography generation of the two waves: the b-wave being due to a fear of injections). the more proximaL However, these findings Methods gave only an approximate indication of sub­ sequent neovascularisation. Other recent Clinical Assessment reports have suggested ERG latencies, which Examination, undertaken by an observer who reflect retinal sensitivity, may be useful addi­ was unaware of the ERG results, included tional indicators of new vessel for­ slit-lamp microscopy, mation.11.12,13 measurement, gonioscopy, fundoscopy and The ERG evoked by a pattern stimulus colour . The following (PERG) is considered to have an even more features were recorded: visual acuity, the proximal retinal origin than the scotopic number of fundal cotton wool spots, the b-wave, 14,15 It might therefore register signs of extent of deep retinal haemorrhages, the a compromised retinal circulation earlier than presence of macular oedema, retino-ciliary a flash evoked ERG. collateral vessels and or retinal new The PERG is a localised test of central ret­ vessels. If present a relative afferent defect inal function and would not reflect the total was quantified using neutral density filters.1 7 extent of hypoxia present in the retina. How­ Fundus Fluorescein Angiography ever Magargal et al. 16 has suggested that broad Fluorescein angiography was performed with zones of capillary nonperfusion extend from a 50° fundus camera taking sequential views of the posterior pole to the retinal periphery. the posterior pole and the mid-periphery of Should this be the case then the PERG, by the retina. Significant ischaemia was defined sampling the posterior pole region, might be as greater than 50% capillary non-perfusion an index of the total amount of retinal ischae­ of the total retinal visible on the fluorescein mia and thus be related to the likelihood of angiogram and/or disruption of the perifoveal subsequent neovascularisation. arcade. The aim of our study was to compare the efficacy of both flash and pattern electroreti­ Electroretinography nography with currently employed clinical PERG recordings are demanding for the and fluorescein angiographic methods as a patient, requiring steady fixation and focus. prognostic indicator of neovascularisation in The total ERG protocol was therefore patients with recent onset CRVO. designed to minimise patient chair time: a recording session including flash and pattern Patients and Methods ERGs typically took 1-1.5 hours. Expendable DTL fibre corneal electrodes in detachable Patients holders18 were referred to EEG cup elec­ Forty-three consecutive patients, who pre­ trodes positioned on the ipsilateral temporal sented to the Birmingham and Midland Eye bone. Binocularly evoked responses were Hospital casualty department between averaged on a Nicolet Pathfinder and stored December 1988-Decmber 1989 with uni­ on magnetic media for subsequent analysis. ocular CRVO, were recruited. Fellow eyes were examined and patients with bilateral Flash ERGs ocular involvement or other major retinal vas- Opaque diffusing goggles were used to distri- 364 A. J. MORRELL ET AL. bute evenly the light flashes emitted by a Patients were followed in clinic for a mean Grass Strobe. Photopic ERGs were recorded of 7.8 months (range 4-14 months), with the to flash intensities 2, 4, 8 and 16. 10-20 exception of one patient who died at two responses were averaged between a bandpass months. Eight of 41 patients developed of 0.5-250Hz. Oscillatory potentials were fil­ rubeosis iridis within a mean of 10.4 weeks tered offline, between ISO-250Hz, from the (range 3-20 weeks) from the estimated date of photopic response to flash intensity 16. A onset of CRVO. Three of these eight patients 30Hz flicker response was elicited by flash also developed new vessels on the disc. For intensity 8 and 20 averages recorded between subsequent analysis the patients were divided a bandpass of 0.5-70Hz. Single flash scotopic into rubeotic and non-rubeotic groups. The ERGs were obtained after a minimum of 15 initial ERG data and clinical findings were minutes dark adaptation to flash intensity 8. compared across the two groups to find the

tTwo patients were unable to tolerate the dark most sensitive predictive factor of rubeosis . adaptation period because of claustrophobia even though the subject room was sizeable Clinical Assessment Results and a bowl stimulus was not used]. Relative afferent pupil defect All patients were examined for the presence Pattern ERGs of a relative afferent pupil defect (RAPD). Four black and white chequerboard patterns Twenty patients were found to exhibit a were used with decreasing effective retinal defect, including all eight rubeotic patients. contrast. They were back-projected onto a The RAPD was quantifiedin 16 patients, but circular screen of mean luminance 250 cd/m2• not in four of the patients who subsequently Stimulus contrast was 75%. The patterns became rubeotic. The presence and magni­ were presented in order of decreasing retinal tude of the RAPD was a sensitive method of contrast: differentiating the two groups of patients The first two patterns were counterphased at (Mann-Whitney test, P<0.0005). 4Hz: 1. A chequerboard subtending 5° 30' Visual acuity 2. A chequerboard subtending 30' In order to include measures of PL, HM and The remaining two patterns were presented CF, visual acuities were grouped into three in onset/offset mode, appearing for 125msec bands: «6/60), (6/60-6/18) and (>6/18). As and then disappearing into a background of shown in Table I, poor visual acuity was found mean luminance for 125msec: to be of significance in distinguishing the 3. A chequerboard subtending 5°30' rubeotic group at the level p<0.025 (Chi­ 4. A radial chequerboard of 112 elements square test). If the visual acuities were arranged around a central circle of 2°30' grouped into two bands: <6/60 and 6/60 or containing a uniform chequerboard of better, then the significancelevel was found to 3'. increase to -<0.015. Results Retinal heamorrhages Four patients were considered to be at risk of The extent· of deep retinal haemorrhages, neovascularisation on the basis of their clini­ graded 0-3 according to the proportion of ret- cal and fundus fluorescein angiographic find­ ings at their initial MOC visit and received Table I Visual acuity. Poor visual acuity was of immediate prophylactic laser photocoagula­ significancein predicting rubeosis at the level p <0.025 tion. Two of the four were treated before elec­ (Chi square test). Grouping the visual acuities into 2 troretinography could be performed and were bands: <6160 and 6160 or better. increased the level of significance to p excluded from this study. For statistical pur­ <0.015

poses the remaining two patients who under­ <6160 616;-6118 >6118 went ERGs before treatment were considered to be part of the non-rubeotic Rubeotic 5 1 2 = 8 group. The finalsample size was therefore 41 Non-rubeotic 6 20 7 = 33 Total 11 21 9 = 41 patients. ELECTRORETINOGRAPHY AS A PROGNOSTIC INDICATOR OR NEOVASCULARISATIONIN CRVO 365

Table II Results of assessment by fundus fluoresceinangiography. Three out of eight patients were identified as being ischaemic and therefore at risk of rubeosis by fluorescein angiography

Classification by fluorescein angiography ischaemic non-ischaemic borderline

Final outcome Rubeotic 3 2 o Prophylactic laser 2 o o Non-rubeotic o 25 2 Total 34

Table III Interocular differencein flicker ERG latency. The phase delay was significantly greater in the rubeotic group: Student t-test , p

Interocular difference in Number of patients flicker latency

Rubeotic 7 9. 78 ms SD = 3.28 Non-rubeotic 31 4.66 ms SD = 3.78

ina involved (0 = absent, 1 = mild, 2 = mod­ Electroretinographic Results erate, 3 = severe), was found to be of Flash electroretinography borderline significance in distinguishing the The flashelectroretinography data in the form two groups (Mann-Whitney significant at of the amplitude ratios e.g. photopic and sco­ p<0.049). topic a:b -wave amplitude, b-wave latency and flickerphase difference between affected Cotton wool spots . and unaffected eyes were statistically eval­ Statistically the number of cotton wool spots, uated. The interocular difference in the grouped (0-3), (4--7), (8-11) and (12-15), was flickerERG latency was found to be the most a poor indicator of subsequent rubeosis significant distinguishing feature (see Table (Mann-Whitney p

Table IV Interocular difference scotopic alb wave when amplitude measures are compared9•1o.13 amplitude ratio. The mean interocular difference in the whilst others have found the latency results scotopic alb wave amplitude was found to be a more accurately distinguish the two groups of significantindicator of rubeosis (student t-test p

Rubeotic 7 0.311 SD = 0.354 both scotopic and photopic a:b amplitude Non-rubeotic 29 0.003 SD = 0.137 ratios to be significant indicators of sub­ sequent rubeosis. Of these, interocular flicker that this number was found to be a sensitive latency difference is the most meaningful dis­ indicator of the likelihood of neovascular­ criminant measure: a difference of six milli­ isation (Table V, Mann-Whitney test seconds or greater would have correctly p

10 one false positive case. The number of cotton wool spots present at initial examination was found to be a poor indicator of subsequent rubeosis whilst retinal haemorrhages and • RUBEOTIC m NON·RUBEonc visual acuity were of marginal significance in I predicting outcome. This data has to be inter­ preted with caution because of the small sample size involved. It is possible that factors

Number Of patterns of apparent marginal significancemay assume greater significance if a larger sample size was Fig. 1. Perg data. 4 patterns of decreasing effective involved. retinal contrast were used to elicit the PERG's. The number of patterns to which a PERG could be recorded Fluorescein angiography has been useg was a significant prognostic indicator of rubeosis at the extensively to assess the degree of retinal p

Statistical analyses were performed with the MINITAB and GUM packages. This comprised the Chi-squar� test for categorical data and the student t-test or Mann-Whitney U test for continuous data group comparisons' The nonparametric test being enlarged if the data were non-symmetrically distributed or inexact (e.g. visuat acuities: CF

Table V Perg data. The number of different patterns to which a response could be elicited was a significant prognostic indicator of rubeosis. Mann-Whitney test p

number of patients median number of patterns

Rubeotic group 7 0.0 Non-rubeotic group 32 2.0 invasive procedure with recognised problems electroretinography-Pattern electroretinography­ of standardising the assessment. The experi­ Fluorescein angiography-Rubeosis. ence of this study confirmed the difficulties References encountered by other authors with either fail­ 1 Magargal LE, Brown GC, Augsburger JJ, Parrish ure to obtain an angiogram in every patient or RK: Neovascular glaucoma following central ret­ inal vein obstruction. 1981; 88: of technical problems of interpretation e.g. 1095- 100. due to masking by retinal haemorrhage. 19 For 2 Hayreh SS: Classification of Central Retinal Vein such reasons only three of eight rubeotic Occlusion. Ophthalmology 1983; 90: 45 8- 72. patients in this sample group were success­ 3 Laatikainen L and Kohner EM: Fluorescein angiog­ raphy and its prognostic significance in retinal fully identified as being at risk by fluorescein vein occlusion. Br J Ophthalmol1976; 60: 411-18. angiography. 4 May DR, Klein ML, Peyman GA, Rauchand M: Although criteria such as depth of RAPD Xenon arc panretinal photocoagulation for cen­ and ERG measures are sensitive indicators of tral retinal vein occlusion: a randomised prospec­ tive study. Br J Ophthalmol1979; 63: 725- 34. rubeosis, the false positive rate using multiple 5 Laatkainen L: Preliminary report on effect of retinal discriminant analysis combining both clinical panphotocoagulation on rubeosis iridis and neo­ and ERG data would improve the specificity. vascular glaucoma. Br J Ophthalmol 1977; 61: We are currently examining its possibility. 278- 84. 6 Katz A: Clinical and experimental studies on retinal Conclusions neovascularisation. Am J Ophthalmol 1982; 94: In this preliminary study significant distin­ 71 5- 43. 7 Magargal LE, Brown GC, Augsburger JJ, Donoso guishing features with high predictive values LA: Efficacy of panretinal photocoagulation in for the development of rubeosis iridis are:­ preventing neovascular glaucoma following (1) the interocular flickerlatency difference ischaemic central retinal vein occlusion. Ophthal­ (2) the ability to elicit pattern ERGs mology 1982; 89: 780- 4. 8 Welch JC and Augsburger JJ: Assessment of angio­ (3) the depth of the relative afferent pupil graphic retinal capillary non-perfusion in central defect retinal vein occlusion. Am J Ophthalmol 1987; (4) the ratio of scotopic and photopic a:b 106: 35 H. wave amplitude ratios 9 Sabates R, Hirose T, McMeel JW: Electroretinog­ Factors of lesser statistical prognostic value raphy in the prognosis and classification of central retinal vein occlusion. Arch Ophthalmol, 1983; included: 101: 232-5. (5) a visual acuity less than 6/60 to Barber C, Galloway N, Reacher M, Salem H: The (6) the extent of deep retinal haemorrhages role of the electroretinogram in the management of central retinal vein occlusion. Doc Ophthalmol (7) fundus fluorescein angiography Proc Series 1984; 40: 149-59. (8) the number of cotton wool spots. 11 Johnson MA, Marcus S, Elman MJ, McPhee TJ: Neovascularisation in central retinal vein occlu­ We are grateful to Mr. Peter Francois, Senior Lec­ sion: electroretinographic findings. Arch Oph­ turer, Dept of Vision Sciences, Aston University and thalmol1988; 106: 348-52. Mr. David Shaw, Research Assistant, University of 12 Kaye SB and Harding SP: Early electroretinography Birmingham, Department of Ophthalmology for the in unilateral central retinal vein occlusion as a pre­ statistical analysis performed on the data of this study. dictor of rubeosis iridis. Arch Ophthalmol1988; We would like to thank Dr. Lesley Jones and Miss Viv­ 106: 35 3- 6. aca Vann Der Vliet of the Clinical Neurophysiology 13 Breton ME, Quinn GE, Keene S, Dahmen JC, Unit, Dept of Vision Sciences, Aston University for Brucker AJ: Electroretinogram parameters at theirtechnical assistance with the electroretinography presentation as predictors of rubeosis in central and Mr. Ivan Bradley, Birmingham and Midland Eye retinal vein occlusion patients. Ophthalmology Hospital, for his assistance with the photography. 1989; 96: 1343-52. 14 Lawill T: The bar pattern electroretinogram. Doc KEy WORDS: Central Retinal Vein Occlusion-Flash Ophthalmol1984; 40: 1-10. 368 A. J. MORRELL ET AL.

15 Holder GE: Significance of abnormal pattern elec­ 18 Thompson D A and Drasdo N: An improved method troretinography in anterior visual pathway dys­ for using the DTL fibre in electroretinography. function. Br J Ophthalmol1987; 71: 166--71. Ophthal, Physiol Opt 1987; 7: 31 5-19. 16 Magargal EL, Donoso LA, Sanborn GE: Retinal 19 Servais GE, Thompson HS, Hayreh SS: Relative ischaemia and risk of neovascularisation follow­ afferent pupil defect in central retinal vein occlu­ ing central retinal vein occlusion. Ophthalmology sion Ophthalmology 1986; 93: 301- 3. 1982; 89: 1241-5. 20 Minturn J and Brown GC: Progression of nonis­ 17Thompson DA, Corbett JJ, Cox TA: How to chaemic central retinal vein obstruction to the measure the relative afferent pupil defect. Surv ischaemic type. Ophthalmology 1986; 93: Ophthalmol1981: 26: 39-42. 11 58--62.