articles - university of...

Articles

Entoptic Evaluation of Diabetic Retinopathy

Raymond A. Applegate,* Arthur Bradley,^ WichardA.J. van Heuven* Bailey L. Lee,*and Charles A. Garcia*

Purpose. Studies using optimized entoptic viewing of the parafoveal retinal vasculature haveshown that normal subjects see their own capillaries with greater detail in the fovea than seentypically in fluorescein angiography. The authors have extended these investigations to per-sons with diabetes to evaluate the sensitivity, specificity, and accuracy with which they candetect and locate their own parafoveal retinal defects untrained.

Methods. A vascular entoptoscope using Maxwellian view optics creates a high-contrast entopticview of retinal vasculature abnormalities in the parafoveal area. Using a double-masked proto-col, 70 patients with diabetes and 29 control subjects described, drew, and quantified theirentoptic image. These entoptic records were compared to angiograms and color photographsobtained immediately after the entoptic evaluation.

Results. Angiograms or color photographs or both showed that 61 of 70 patients with diabetes hadretinal defects (e.g., microaneurysms or exudates or both) within the field of view of the VascularEntoptoscope (8.1° or 11.6° circular field depending on the Vascular Entoptoscope used: parafovealarea subtends «9.7°). Of these 61 patients with diabetes, 51% (31) observed dark "spots" or"blobs" in the entoptic field corresponding to retinal defects in the angiograms or photographsor both. Seven (18%) of the 38 patients (9 patients with diabetes and 29 control subjects withoutdefects in the entoptic field) said they saw something when angiograms or photographs or bothshowed nothing (false-positive). Thus, the sensitivity and specificity (using angiograms or photo-graphs or both as the gold standard) with which untrained patients with diabetes detect their ownparafoveal area defects are 51% and 82%, respectively. Superimposition of the entoptic image (asdrawn by the patient) and the angiograms or color photography or both often showed excellentcorrespondence. Most (22 of 29) of the control subjects and more than half (40 of 70) the patientswith diabetes were able to quantify the size of their foveal avascular zone (FAZ) from the entopticview, whereas only 22 of 70 of the capillary loops defining the FAZ were visible in the optimalframe of the capillary phase of the fluorescein angiogram. As reported previously in a smallersample, large FAZs often were associated with poor visual acuity.

Conclusions. More than half the untrained patients with diabetes were able to visualize theirown parafoveal retinopathy entoptically, and most untrained patients with diabetes and con-trol subjects where able to quantify the size of their FAZ. Patients and control subjects withoutparafoveal defects rarely report defects not visible photographically. Patients can be trainedto detect their defects. Clinical entoptic monitoring will require verification that patients candetect changes in their retinopathy. Entoptic testing is low cost, noninvasive, and can beperformed as often as needed at no risk to the patient. It is, therefore, a promising researchtechnique for subjective monitoring of the early natural history of parafoveal area diseaseprocesses. Invest Ophthalmol Vis Sci. 1997;38:783-791.

From the * Department of Ophthalmology, University of Texas, Health Science Center Each year, diabetes Causes 65,000 new Cases of prolif-at San Antonio, San Antonio, Texas; and the f School of Optomtry, Indiana erative eye disease and 8000 new Cases of blindness,UniversityBioomngfm. „ , , „ , , , , , . , , „ . . , , . . . . , m a k i n g d i a b e t i c r e t i n o p a t h y t h e m a j o r c a u s e o f b l i n d -Supported by Grant EY10097 and Supplement EY10097-A1S1 (RAA) from the National ° , r ' 1-3Eye Institute, National Institutes ofHeaUh, Bethesda, Maryland; a grunt fiom the San neSS in the working age population. It is estimatedAntonio (Texas) Ana Fmi?idalion (RAA); and an unrestricted research grant to the fa^ there are 500,000 Cases of macular edema WithDepartment of Ophthalmology, The University of Texas Health Science Center at San M^ „ , . , , n 4 n i- i iAntonio fmm Search to Prevent Blindness, New York. 75,000 new cases annually. Studies have shown thatSubmitted for publication May 22, 1996; revised November 7, 1996; accepted July 15, laser treatment of diabetic retinopathy and, in particu-19% „ lar, of clinically significant macular edema can reduceProprietary interest category: N (WAJV, BLL, CAG); P (RAA, AB). . ' , , , - . . r ,. ,Reprint requests: Raymond A. App'legate, Department of Ophthalmology, Significantly the 1OSS of Vision from diabetic eye dlS-UTHSCSA, 7703 Floyd Curl Drive, San Antmio, TX 78284-6230. ease.4 Because laser treatment is destructive, it is not

Investigative Ophthalmology & Visual Science, April 1997, Vol. 38, No. 5Copyright © Association for Research in Vision and Opluhalmology 783

784 Investigative Ophthalmology & Visual Science, April 1997, Vol. 38, No. 5

initiated until the disease has reached a point wherevision loss is believed to be imminent. Unfortunately,diabetic retinopathy often is asymptomatic during itsmost treatable stages4 and, as a consequence, sight-threatening changes in diabetic retinopathy that oc-cur between regular eye examinations can go unno-ticed. Given the recent advances in laser treatmentand its demonstrated effectiveness, early detectionand monitoring of eye pathology in diabetes has beenidentified as an area where "if the Federal Govern-ment does not pay the cost of preventing blindness(i.e., the cost of detection and monitoring), it will surelypay the price of the disability," which is seven timeshigher.5

Entoptic visualization of the retinal vasculature isachieved by producing a moving high-contrast shadowimage (Fig. I)6 of the vessels in the plane of the photo-receptors. It is important that the shadows move (stabi-lized retinal images are not visible or disappear orboth7), which can be achieved by moving the entrypoint of the light source.

Several techniques have been developed to allowentoptic visualization of the retinal vessels, but as earlyas the nineteenth century, Helmholtz8 observed strik-ing foveal vascular detail when illuminating the retinadirectly through the natural pupil with a small pointsource (the exit pupil of a microscope) movingaround in the eye's pupil plane. In our research, wehave capitalized on the observations by Helmholtz8

and designed a Maxwellian-view instrument: the Vas-cular Entoptoscope. Details of the theory and opera-tion of this device have been presented earlier.9 Stud-ies of normal subjects show that foveal capillary detailinvisible with standard fundus photography and ofteninvisible with fluorescein angiography can be seen us-ing entoptic testing.

Clinically, entoptic visualization has been used tohelp evaluate the functional status of the retina be-hind obstructed media.1011 To our knowledge, otherthan our own studies,12'13 only one study has used en-toptic visualization of the retinal vessels to monitor anactive disease state. Kluxen and Wilden14 trained 136insulin-dependent patients with diabetes to observetheir retinal vasculature entoptically using a transscle-ral illumination technique. They found that in pa-tients with one to five microaneurysms as shown byfluorescein angiography, 55% entoptically could de-tect their own pathology. In patients with 6 to 20 mi-croaneurysms, the percentage increased to 77%. Inpatients with more than 20 microaneurysms with se-vere background and proliferative retinopathy, 90%reliably could detect their own pathology, and manycould document the appearance of new and the disap-pearance of old microaneurysms over time.

Combining the observation of Kluxen andWilden14 with the report by Klein et al,'a'16 showingthat the number of microaneurysms in the diabetic

FIGURE l. Fundus photograph using oblique illumination ofthe fundus. An illuminating fiber optics bundle was appliedon the conjunctiva. The angle of incidence of light at theposterior pole of the eye was estimated to be 30°, and theprojected direction is indicated by an arrow. The distancebetween the vessel and its shadow is 149 ± 10 fim (averageof 16 sites). The shadow is formed on a surface located«140/tan(30°) or 240 ^m posterior to the vessels. (Re-printed with permission of the publisher, The Optical Soci-ety of America, and the first author, Francios Delori. AppliedOptics. 1989;28:1069.6)

eye is correlated directly with the severity of the dia-betic retinal disease, along with two experimental stud-ies17'18 showing that the foveal avascular zone (FAZ)enlarges as the severity of the diabetic retinopathyincreases, and die fact that parafoveal hard exudatesare a result of macular edema, we thought that anoptimized entoptic viewing technique (as opposed tothe cruder transscleral penlight technique) may be anexcellent noninvasive tool for monitoring the naturalhistory of diabetic retinopathy in the macula beforevisual acuity loss.

Over the past several years, we systematically haveinvestigated the entoptic visualization of the retinalvasculature and its clinical implications.9121319'20 Wehave shown that large vessels over a wide area of the

Entopic Evaluation of Diabetic Retinopathy 785

retina, smaller vessels in the parafoveal region, andcapillaries in the fovea can be observed readily usingoptimized entoptic viewing procedures.20

In applying entoptic methods to the detection andmonitoring of diabetic retinopathy, we have made thestrategic decision to target the parafoveal and fovealarea specifically, because the impact of diabetic reti-nopathy on vision is most profound when the patho-logic changes affect those areas. These are also theexact areas in which entoptic viewing has its highestresolution.919

In unpublished pilot work, we showed, similar tothat of Kluxen and Wilden,14 that we could train pa-tients with diabetes to detect their own parafoveal areamicroaneurysms, hard exudates, and retinal hemor-rhages. Encouraged by these findings and the nonin-vasive nature of the test, we believed that entopticviewing of the central retinal area could provide theclinical researcher with fundamental information con-cerning the early natural history and pathogen esis ofparafoveal area disease and offer the potential forearly detection and diagnosis, a firm foundation onwhich to base a rationale for new preventative therapy,and a sensitive means of evaluating therapy and treat-ment designed to alter the natural course of the dis-ease.9

Here we report results of an ongoing NIH- andNEI-supported study designed to evaluate the sensitiv-ity and specificity with which untrained diabetics canuse the optimized entoptic viewing technique to de-tect, describe, and quantify their own parafoveal pa-thology.

METHODSSubjectsSeventy patients with diabetes ranging in age from 18to 70 years and 29 age-matched normal control sub-jects visiting the Ophthalmology Clinic of the Univer-sity Health Center, San Antonio, were asked to serve assubjects. Age-matched control subjects were recruitedfrom patients with fundus photography of abnormaloptic papillae or choroidal nevi and had no observableparafoveal pathology clinically or photographically.Although no subjects were excluded or selected basedon race, the majority (78) were Hispanic, which re-flects closely the racial makeup of our clinic.

Classification of disease severity was determinedby the Wisconsin Fundus Photographic Reading Cen-ter using the Modified Airlie House Classification ofDiabetic Retinopathy as adapted for the Early Treat-ment of Diabetic Retinopathy Study.2122 The distribu-tion of disease severity as classified by the readingcenter is displayed in Figure 2.

Entoptic InstrumentationA Vascular Entoptoscope identical in principle to thatdescribed by Applegate et al9 modified for the clinical

3 U

(0 25

l »O<5 1 0

1 50

iW

m NOrmals

^ ^ ^ ^ H H H Diabetics

• •Ifc0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 0 1 0 0

Modified Airlie HouseClassification as adapted

by ETDRS

FIGURE 2. Frequency distribution of patients as a function ofthe Modified Airlie House Classification as adapted by theEarly Treatment of Diabetic Retinopathy Study.

environment13 was used to collect the entoptic data.The Vascular Entoptoscope uses Maxwellian-view op-tics to image a small (1-mm diameter) short-wave-length (blue peak wavelength X = 430 nm) pointsource in the subject's pupil plane, where it travels ina circular path (diameter, 4 mm) at 3 to 5 Hz andimages an 8.1° or 11.6° (depending on the particularinstrument) diameter blue field onto retina. To mea-sure FAZ size, the circular field size is adjustable (Fig.3) down to ^0.3° using either an electronically con-trolled aperture or a manual control knob (dependingon clinical instrument used). A small dim fixationlight was located in the field center to aid in subjectfixation and locating parafoveal defects with respectto the retinal locus of fixation.

Experimental Procedures

Tenets of the Declaration of Helsinki were followed,Institutional Review Board approval was obtained, andsigned informed consent was obtained from each pa-tient. Subjects viewed a video showing how the deviceworked and what they were going to be asked to do.Neither the subject nor the research assistant testingknew if the patient in fact had foveal or parafovealarea defects. Therefore, unlike previous work, patientswere not trained to see their own defects using feed-back from angiograms or color photography.

After viewing the video, patients observed theirfoveal and parafoveal area entoptically using the Vas-cular Entoptoscope and drew their FAZ and any irreg-ularities in the vessel pattern, such as spots and blobs.

If spots or blobs were seen, the size of each defectwas recorded using an eight-level scale provided onthe patient drawing sheet, and the location of thedefect was drawn within the field of view with respect


FIGURE 3. A high-contrastdrawing made from an his-tologic preparation illus-trating a normal foveal avas-cular zone (FAZ) and theinstrument aperture stop asit appeared to our subjects.The detailed vascular pat-tern is similar to that ob-served by subjects. The FAZdiameter was measured byhaving each subject de-crease the size of the fieldstop until it matches theFAZ border. The subject'sview during this procedureis shown schematically inthis series of pictures as thefield stop is decreased insize. (Reprinted with per-mission of the publisher,Butterworth - Heinemann,and the first author, ArthurBradley. Ophthalmic PhysiolOpt. 1992;12:20.20)

to fixation point. For example, data from the patientwhose fluorescein angiogram (FA) is displayed in Fig-ure 4A saw four spots. Two of the spots were scaledas size 1, and two of the spots as size 2 (Fig. 5). Incomparing the drawing (Fig. 5) to the FA in Figure4A, remember the entoptic field of view is invertedalong the horizontal meridian through fixation withrespect to the FA.

After drawing any visible defects, patients wereasked to decrease the field stop diameter until theedge of the stop denned the edge of the FAZ (Fig. 3).The angle the field stop subtended at the eye's en-trance pupil (x) served as the quantitative measure ofFAZ diameter and could be converted easily to mi-crometers in the plane of the retina by assuming thatthe eye's posterior nodal distance was 16,670 fj,m (FAZ= tan(x)-16,670).

Immediately after entoptic testing, routine colorfundus photography and fluorescein angiographywere obtained for comparison with the entoptic data.FAs were performed only on the patients with diabe-tes.

Photographic Imaging of the Fundus

A Zeiss FF-3 30° fundus camera (Zeiss Strasse 73446;Carl Zeiss, Oberkochen, Germany) was used to obtain

color fundus photographs and perform the fluores-cein angiography. Care was taken by the retinal pho-tographer to obtain the best photodocumentadon ofthe parafovea during the capillary phase of the FA tooptimize the visibility of the capillaries defining theFAZ.

Image Processing

Fundus images (color and FA) were digitized andstored on photograph computer disks, and the entop-tic drawings were scanned and digitized using anApple scanner and a Macintosh computer (Apple, Cu-pertino, CA). The digitized entoptic images then weretraced using a standard object-based graphic package(MacDraw; Claris, Santa Clara, CA) to create high-contrast images with transparent backgrounds, whichthen were superimposed onto the fundus images forcomparison.

To compare the photographic and angiographicimages with the entoptic drawings, traced entopticdrawings were superimposed onto the photographicimages. Before super-position, the entoptic imageshad to be top-bottom inverted to convert the per-sonal view to the same coordinates as that of the clini-cal view. Scaling was achieved by following a standardrule for each instrument. For example, the circular


field limit to the entoptic image (for 8.1° instrument)was scaled to equal 52% of the distance from the fove-ola to disk center (15.5°) and centered on the pre-sumed foveal center.

RESULTS

Of the 70 patients with diabetes whose eyes we tested,61 (87%) had angiographically or photographicallyvisible pathology or both within the central 8.1° or10.1° (depending on the instrument used) field. Ofthese 61 patients with diabetes, 31 (50.8%) reportseeing either spots, blobs, or other abnormalities inthe entoptic image. That is, the entoptic method wassuccessful in detecting 51% of the patients with pathol-ogy within the field of view. Thirty-eight patients (9patients with diabetes and 29 control subjects) had nodefects within the entoptic field of view. Of these 38patients, 7 reported seeing parafoveal area defects not

B

FIGURE 4. Examples from two different patients (A, B) show-ing the super-position of the entoptic image (white lines andcircles) onto a negative view of the capillary phase of themacular fluorescein angiogram. The dashed circles indicatethe entoptic field of view. The small circles are entopticallyvisible dark "spots" or "blobs." The large irregular shapein the center of Figure 4A is the patient's drawing of thecapillary arcade forming the foveal avascular zone.

FIGURE 5. Entoptic drawing of a patient with diabetes whosefluorescein angiogram (FA) is displayed in Figure 4A. Thepatient saw entoptic images of four prereceptor retinal de-fects within the field of view. The locations of the defectwith respect to fixation and the edge of the field were drawnon a sheet of paper and the size of the defect labeled usingan eight-level scale. In this case, the patient labeled thedefects as size 1 or size 2. The entoptic drawing is invertedalong the horizontal meridian through fixation with respectto the FA (Fig. 4A).

visible with photographic documentation, or an 18%false-positive rate. We can calculate from Table 1 thatthe entoptic method, when used by untrained pa-tients, has a sensitivity of 51% and a specificity of 82%.

Simply because untrained patients can detecttheir retinal pathology 51% of the time does not indi-cate necessarily that the entoptic method provides anaccurate image of the pathology. To evaluate the accu-racy with which patients with diabetes can locate reti-nal defects, we used a super-position approach. Thatis, we scale and superimpose the patients' entopticdrawings onto digital images of their FAs or field 2

TABLE l. Entoptoscopic Reports

Patients with diabetes withphotographically visibledisease in the entoptic field

Controls and patients withdiabetes with nophotographically visibledisease in the entoptic field

DetectDisease

31

7

Fail toDetectDisease

30

31


FIGURE 6. Examples from two additional patients showingthe super-position of entoptic and angiographic images ofthe central retina.

of their color photographs or both. Examples of thisprocedure can be seen in Figures 4, 6, and 7 fromsubjects where both entoptic data and FAs were ofgood quality.

Figure 4 shows examples from two different pa-tients (Figs 4A, 4B) in whom the macular region ofthe angiogram is shown as a photographic negative.We have chosen to use the negative because this is theview seen entoptically by patients where vessel shadowsand opaque pathology appear dark on an otherwiselight (blue) field. Superimposed onto the angiogramsin white are the entoptic drawings made by the pa-tients, which include numerous spot opacities drawnas small circles. The dashed white circle describes theentoptoscope field of view, and the large irregularshape in the center (Fig. 4A) is the FAZ traced by thepatient.

Figure 4A shows the angiogram from a diabeticright eye (scored 43 by the reading center) with sev-eral obvious microaneurysms (MAs) surrounding thefoveal center. Three MAs, one directly superior, oneinferior, and one directly temporal to the foveal cen-ter, can be seen easily. There are several other moreperipheral MAs inferior and temporal to the fovealcenter and more outside of the entoptic field of view.

The entoptic drawing made by this patient shows acentral avascular zone (irregular shape in the circlecenter) and four visible spots: one superior, one infe-rior, one temporal, and one inferior-temporal. Thelocations of the drawn MAs do not correspond per-fectly with those of the angiographic image, but theyare close. This patient has seen as much pathology inthe parafovea as the clinician might see.

Figure 4B shows an FA from the left eye of a differ-ent patient with numerous aneurysms just nasal anddirectly inferior to the foveal center. Entoptically, thisis close to what the patient reports and draws. Thispatient drew a large blob with eight smaller dots dis-tributed in a horizontal band. When superimposed,the spatial correspondence between the entopticdrawing and the angiogram is quite good.

Figures 6A and 6B show two more examples inwhich retinopathy can be seen clearly in the angio-gram, and the patient in each case is able to reportseeing opacities with similar spatial distributions. Fig-ure 6A shows an FA with a large MA just inferior andtemporal to the foveal center, and the patient drawstwo spots close to this location. The FA also shows twoclusters of small aneurysms superonasal and supero-temporal to the fovea. The patient has drawn a pairof spots in both locations. The left eye of a differentpatient (Fig. 6B) has retinopathy visible superior andalso temporal to the foveal center, but the patient onlydraws pathology superior near where there are twoeasily visible microaneurysms.

During the pilot study for this project, we testedseveral subjects who were selected specifically becausewe knew that they had foveal retinopathy. Althoughnot included in any of the data analysis described inthis report, we have included a super-position imagefrom one of these patients (Fig. 7) because it shows a

FIGURE 7. Super-position of entoptic (white) and angio-graphic images from the left eye of a patient with diabetes(trained to see his or her own defects using the fluoresceinangiogram) with numerous microaneurysms in die foveaand parafovea.


w^ 1

P o

If0 TJ

< 0LJ.Q

2000

1500

1000

500

D DD

0 K--0 500 1000 1500 2000

FAZ diameter (microns)Determined Angiographically

FIGURE 8. Scattergram of foveal avascular zone diameter ob-tained with entoptic (x) and angiographic (y) methods (di-agonal line is the y = x line).

patient can be trained to see the foveal pathology.The left eye of this patient has a clear arc of MAs justtemporal to the central fovea, an obvious MAjust nasalto the foveal center, and several MAs near to the infe-rior edge of the entoptic field. Entoptically, this pa-tient draws an arc of spots just nasal to the fixationpoint and a visible spot just temporal and near to theinferior edge of the entoptic field. When superim-posed, the angiographic and entoptic images showa clear similarity, indicating that this patient had anentoptic view of the pathology similar to that availableto the clinician.

Size of Foveal Avascular Zone

It has been reported that in persons with diabetes,FAZ size increases and can become irregular inshape.1718 Because the capillary arcade forming theFAZ is one of the most striking characteristics of theentoptic foveal image in normal persons9'1319'20 andcan be seen entoptically by most normal persons, wewere motivated to see if patients with diabetes, wereable to report their FAZ size correctly. To comparethe entoptic estimates of FAZ diameter with the angi-ographic estimates, a quantifiable FAZ is required inboth the angiographic and entoptic images. Forty ofthe patients with diabetes were able to measure reli-ably their FAZ entoptically. Of these 40, 22 had FAswith sufficient foveal capillary detail to quantify FAZsize. To examine the relation between the entopti-cally determined and the angiographically deter-mined FAZ size, we have plotted the two measuresof FAZ diameter in Figure 8.

As shown in Figure 8, most of the 22 subjectsreport seeing an FAZ that is similar in size to thatobserved angiographically. Using entoptic visualiza-tion, three subjects report an FAZ diameter consider-ably larger than that seen angiographically, and one

subject reports an FAZ much smaller than seen angio-graphically.

In an earlier sample of patients with diabetes,12

we reported a decrease in high-contrast visual acuityas entoptically measured FAZ diameter increases. Weexamined the relation again here now that the sam-ple size has increased by more than a factor of 5.Although there is considerable scatter (Fig. 9), thosesubjects with enlarged FAZs tend to have worse acuity(P< 0.001, r = 0.53).

DISCUSSION

Optimized entoptic viewing of the parafoveal reti-nal pathology and vasculature can be performedby untrained patients with diabetes. It is noninva-sive and can be implemented as often as needed.

Previous experience with the Vascular Entopto-scope has shown that foveal capillary detail, espe-cially the capillary arcade forming the FAZ, is visi-ble to virtually 100% of normal persons. One dem-onstration of the ease with which persons can seefoveal area capillary detail came at the annualmeeting of the Optical Society of America in 1990,where we estimate that 400 attendees used the Vas-cular Entoptoscope and all but one (an older manwith small pupils) were able to see their foveal arearetinal vasculature and FAZ. This qualitative find-ing is consistent with our experimental findings onnormal persons.9'19'20 Further, we have shown thatthe smallest capillary detail was visible entopticallyin the fovea and parafovea by taking FAs repeatedlyon ourselves until we were able to get angiographicimages of sufficient quality to see the inner capil-lary loops of the FAZ.

Given the success with educated normal per-

0.8 T

0 500 1000 1500 2000FAZ diameter (microns)

FIGURE 9. Best-corrected high-contrast visual acuity ex-pressed as the log of the minimum angle of resolution as afunction of foveal avascular zone diameter determined usingentoptic techniques (squares) and fluorescein angiogram(circles).


sons, the failure of 49% of the patients with diabe-tes to detect their own parafoveal pathology entop-tically was puzzling at first. There are at least threepossible reasons for the failure of these patientswith diabetes to see their own foveal or parafovealpathology. First, they may fail to see their pathologybecause of reduced visual function caused by reti-nopathy. However, this is unlikely because most ofour patients had good acuity (20/30), with a meanlog of the minimum angle of resolution of 0.18.Second, it is likely that the difficulty of the task(i.e., seeing, locating, and drawing complex retinalfeatures that cannot be fixated) may have been acontributor to these failures. Third, it is possiblethat more complex psychological reasons inter-fered with patient performance. For example, thelow level of education and motivation of our clinicpopulation (for this patient group, the averageyears of education is 10.6 and has a "no-show" ratefor appointments of 50%).

We expect to find a higher sensitivity when test-ing patients who are better educated (the averageyears of education for the Optical Society Americanmeeting demonstration probably was 20 or moreyears) and more motivated. Pilot work in our pri-vate-paying Consultants Clinic supports this belief,as does an analysis of our data set by level of educa-tion. That is, when we compared the retinopathyrating (from the reading center) of those who sawspots on the entoptoscope and those who did not,we found the following:

1. In all patients who saw spots (no selection foreducation level), there was a mean differencefrom those who did not see spots of 2.69 (P =0.5).

2. For high school graduates who saw spots, therewas a mean difference from those high schoolgraduates who did not see spots of 9.84 (P =0.08).

These results support our clinical impression byshowing that patients with high school educationswere more likely to see their retinal defects than wereless-educated patients.

We are particularly encouraged by the fact thateven in this difficult patient population, capillary de-tail of the FAZ was quantifiable entoptically in 40 of70 patients compared to 22 of 70 from the optimalframe of the capillary phase of the FA. That is, thenumber of entoptically visible FAZs was 1.8 times thenumber of FAZs visible in the FAs. This finding is notnew. Bird and Weale,23 our group,91920 and others24

have noted that even very good FAs miss a lot of capil-lary detail.

Of the 22 eyes in which we were able to quantifythe FAZ both entoptically and with FA, most reportan FAZ seen entoptically that matches approximately

that seen angiographically (Fig. 8). Three patients re-ported larger FAZs entoptically and one smaller. Al-though we do not know the exact reasons for theseerrors, we suspect three possible causes. First, as Wein-haus et al24 have shown, even on excellent angiograms,many capillaries are not visible. Compound this withthe fact that typical angiograms on diabetics, wherethe media may not be ideal, foveal capillaries may ormay not be visible and, if they are, what one believesto be the capillary loop defining the FAZ may or maynot actually be the denning loop. Second, given thatmany of our patients with diabetes were unable toreport seeing highly visible pathology in the fovea, itis not surprising that some would not see some of theinner-most capillary loops forming the FAZ. There-fore, we might expect overestimates of the FAZ sizein some of our patients with diabetes. Third, becausenaive subjects generally try to fixate the retinal detailsto which they are attending, the FAZ will always tendto appear larger because its edge will move away fromthe entoptic field center every time the patient triesto fixate the FAZ edge.

The fact that many patients (see five examples inFigs. 4, 6, and 7) can detect and locate and judgeaccurately the size of foveal and parafoveal MAs isencouraging. In fact, a sensitivity result of 51% and aspecificity result of 82% are similar to other clinicaltest results that have had years of development andrefinement. For example, in 182 eyes, it was foundthat the Henson perimeter had a sensitivity of 59.4%and a specificity of 88%, the Humphery perimeter asensitivity of 64.2% and a specificity of 64.2%, and thePerikon perimeter a sensitivity of 55% and a specificityof 90.4% for screening for glaucomatous field de-fects.25 Similarly, it has been found that applanationtonometry alone has a sensitivity of 47.1% and a speci-ficity of 92.4% for screening for glaucoma.26

Our findings confirm our theoretical expectationsthat MAs, exudates, and other opacities that are largerthan are the capillaries should be visible in the parafo-veal area. Combine this with a high specificity of 82%and we reasonably can be assured that untrained pa-tients who see abnormalities entoptically in their para-fovea warrant further clinical evaluation. That is, ifuntrained patients say tiiey see something, then theyprobably do (high specificity). However, given a sensi-tivity of 51%, we can not be comfortable saying thatfurdier testing is not warranted if untrained patientsdo not see defects.

In monitoring patients with known abnormalities,sensitivity can be improved significantly by trainingpatients to see their own parafoveal defects using feed-back. Thus, entoptic testing may prove to be valuableparticularly for monitoring motivated patients be-tween routine examinations to gain a better under-standing of the natural history of the disease process.

In summary, we have shown, in a double-masked


experimental paradigm, that untrained patients withdiabetes can detect entoptically their own parafovealretinal defects with a sensitivity of 51 % and a specificityof 82%. Prior work,14 as well as unpublished work inour laboratory, shows that sensitivity can be increasedwith feedback training. A sensitivity of 51% with aspecificity of 82% is not ideal. These levels of sensitivityand specificity are encouraging given the nature ofour test population and the fact that this is our firstclinical attempt at using the device, and they are simi-lar to accepted well-developed clinical tests for glau-coma screening. We also have shown that FAZ detailwas seen more often entoptically (1.8 times) than wasangiographically. These observations reinforce our be-lief that entoptic monitoring of the parafoveal area byat-risk patients can be an effective, low-cost methodfor documenting the natural history of diabetes beforea visual acuity loss as well as for monitoring therapiesdesigned to alter the disease course.

Key Words

diabetic retinopathy, entoptic images, fluorescein angiogra-phy, microaneurysms, visual acuity

Acknowledgments

The authors thank the subjects and the staff of the Ophthal-mology Clinics for their time and effort, particularly JanetJ. Hendricks, who worked with the patients and collectedthe majority of the data.

References

1. Cotton P. Advances in diabetic retinopathy could savesight, money with more frequent eye exams. JAMA.1990; 264:2608-2611.

2. Kahn HA, Moorehead HB. Statistics on blindness inthe model reporting area 1969-1970. DHEWPublNo.(NIH) 73-427. 1973.

3. National Society to Prevent Blindness Operational Re-search Department. Vision Problems in the U.S.: A Statis-tical Analysis. New York: National Society to PreventBlindness; 1980:1-46.

4. Patz A, Smith R. The ETDRS and Diabetes 2000. Oph-thalmology. 1991;98:739-740.

5. JavittJC, Canner JK, Sommer A. Cost effectiveness ofcurrent approaches to the control of retinopathy inType I diabetics. Ophthalmology. 1989;96:255-264.

6. Delori F, Pflibsen K. Spectral reflectance of the humanocular fundus. Applied Optics. 1989; 28:1061-1077.

7. Ditchburn R. What is psychophysically perfect im-age stabilization? Do perfectly stabilized images al-ways disappear?: Comment. / Opt Soc Am A. 1987;4:405-406.

8. Helmholtz H. In: Southall JPC, ed. Treatise on PhysiologicalOptics. New York: Dover Publications; 1962:217-218.

9. Applegate RA, Bradley A, van Heuven WAJ. Entopticvisualization of the retinal vasculature near fixation.Invest Ophthalmol Vis Sci. 1990;31:2088-2098.

10. Duke-Elder S, Smith RJH. Entoptic Methods. In:

Duke-Elder S, ed. The Foundations of Ophthalmology,Heredity, Pathology, Diagnosis, and Therapeutics. St.Louis: Mosby; 1962:454-458.

11. Murrillo-Lopez F, Maumenee A, Guyton D. Percep-tion of Purkinje vessel shadows and foveal granularpattern as a measure of potential visual acuity. ARVOAbstracts. Invest Ophthalmol Vis Sci. 1993; 34:1422.

12. Bradley A, Applegate R, van Heuven W, Nair P. FAZenlargement and visual acuity in diabetic retinopathy.ARVO Abstracts. Invest Ophthalmol Vis Sci. 1994;35:1395.

13. Nunez R, Applegate R, Bradley A, Hendricks J. Clini-cal version of the Vascular Entoptoscope. Vis Sci ApplTechDigest (Optical Society of America). 1996;l:160-163.

14. Kluxen G, Wilden E. An entoptic test in diabetic pa-tients. Diabetes Care. 1987; 10:800-801.

15. Klein R, Meuer S, Moss S, Klein B. The relationshipbetween microaneurysm count to the 4-year progres-sion of diabetic retinopathy. Arch Ophthalmol.1989; 107:1780-1785.

16. Klein R, Meuer S, Moss S, Klein B. Retinal mi-croaneurysm counts and 10-year progression of di-abetic retinopathy. Arch Ophthalmol. 1995; 113:1386-1391.

17. Bresnick GH, Condit R, Syrjala S, Palta M, Groo A, KorthK. Abnormalities of the foveal avascular zone in diabeticretinopathy. Arch Ophthalmol. 1984; 102:1286-1293.

18. Mansour M, Schachat A, Bodiford G, Haymond R.Foveal avasculature zone in diabetes mellitus. Retina.1993; 13:125-128.

19. Zeffren BS, Applegate RA, Bradley A, van HeuvenWAJ. Retinal fixation point location in the foveal avas-cular zone. Invest Ophthalmol Vis Sci. 1990; 31:2099-2105.

20. Bradley A, Applegate RA, Zeffren BS, van HeuvenWAJ. Psychophysical measurement of the size andshape of the human foveal avascular zone. OphthalmicPhysiol Opt. 1992; 12:18-23.

21. Early Treatment of Diabetic Retinopathy Study Re-search Group. Grading diabetic retinopathy from ste-reoscopic fundus photographs—an extension of theModified Airlie House Classification. Ophthalmology.1991;98:786-806.

22. Early Treatment of Diabetic Retinopathy Study Re-search Group. Fundus photographic risk factors forprogression of diabetic retinopathy. Ophthalmology.1991;98:823-833.

23. Bird AC, Weale RA. On the retinal vasculature of thehuman fovea. Exp Eye Res. 1974; 19:409-417.

24. Weinhaus RS, Burke JM, Delori FC, Snodderly DM.Comparison of Fluorescein Angiography with micro-vascular anatomy of Macaque retinas. Exp Eye Res.1995;61:1-16.

25. Marraffa M, Marchini R, Albertini R, Bonomi L. Com-parison of different screening methods for the detec-tion of visual field defects in early glaucoma. Int Oph-thalmol. 1989; 13:43-45.

26. Tielsch JM, Katz J, Singh K, et al. A population-based evaluation of glaucoma screening: The Balti-more eye survey. Am J Epidemiol. 1991; 134:1102-1110.

articles - university of...

Documents