Studies on the technique of vitreous fluorophotometry Ran C. Zeimer, Jose G. Cunha-Vaz, and Mark E. Johnson We examined, the sources of artifacts of vitreous fluorophotometry (VF), a technique used to assess the permeahility of the hlood-retinal harrier. By performing in vitro measurements in a model eye and in vivo readings in humans, we investigated the influence of various concentrations offluorescein in the anterior chamber and in the retina on the measurements performed in the vitreous. Upper limits for the artifactual effects are set under clinical conditions, and a way to account for them is proposed. Special attention is given to depth resolution, which was investigated as a function of the region being scanned, the beam width, and the probe size. Further, preliminary results indicate that a procedure to account for light attenuation by the lens is feasible. It appears that reliable quantitative results can be obtained by VF, provided, that the artifacts are acknowledged, minimized, and corrected. (INVEST OPHTHALMOL VIS Sci 22:668-674, 1982.) Keywords: vitreous fluorophotometry, artifacts, blood-retinal barrier, fluorescein

itreous fluorophotometry (VF) is a technique developed for measuring low concentrations of fluorescein in the vitreous body, thereby allowing the permeability of the blood-retinal barrier to be assessed and intraocular fluid dynamics to be studied. 1 ' 2 In patients with diabetes mellitus, hypertension, retinitis pigmentosa, and optic neuritis, VF demonstrates an abnormal penetration of intravenous fluorescein into the vitreous. In diabetic patients this occurs before there is angiographically demonstrable retinopathy From the University of Illinois Hospital Eye and Ear Infirmary, Chicago. This study was supported in part by U.S.P.H.S. grants EY 1792, EY 7032 (Dr. Zeimer), and EY 3106 (Dr. Cunha- Vaz) from the National Eye Institute, National Institutes of Health, Bethesda, Md.; and by a grant from the American Diabetes Association (Dr. Cunha-Vaz). Submitted for publication Aug. 11, 1980. Reprint requests to: Ran C. Zeimer, Ph.D., Department of Ophthalmology, University of Illinois Hospital Eye and Ear Infirmary, 1855 W. Taylor St., Chicago, 111. 60612.

668

and clearly separates diabetic from control populations. 2 " 4 It has been shown that this technique can yield reliable quantitative results. However, it has been pointed out that artifacts can influence the readings of borderline cases. 5 " 7 It is the purpose of the studies presented here to identify the sources of artifacts and to quantitate them under clinical conditions. Materials and methods This study used the VF system previously reported4; the technical data of the apparatus are summarized in Table I. In vitro measurements were performed on a model eye (Fig. 1), which reasonably approximated the optics of an eye with a contact lens. Each compartment was isolated from the adjacent one by a coverglass and could be filled separately with a liquid. The anteroposterior dimensions of the anterior chamber, lens, and vitreous humor were similar to those of a human eye. Since the area illuminated and the volume being sampled were very small in comparison with the dimensions of the eye, the measurements were not influenced by the curvature of the vitreal cavity.

0146-0404/82/050668+07$00.70/0 © 1982 Assoc. for Res. in Vis. and Ophthal., Inc.

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933334/ on 01/26/2017

Volume 22 Number 5

Vitreous fiuorophotometnj 669

Table I. Technical data of the apparatus Slit-lamp microscope Magnification at probe plane Illumination Illumination to detection angle Filters Probes Pnotomultiplier Radiometer Dark current noise

MODEL EYE

Haag-Streit No. 360 x2 V. Mueller fiberoptic illuminator 16° Spectrotech SE4OR and SB5OR Gamma Scientific—450 yum and 150 /xm Hamamatsu R375 Gamma Scientific No. 2900 3 mV at 1200 V a t 0.1 sec time constant

Therefore the approximation of the eye by a cylindrical symmetry was not significantly restrictive. It will be seen later that the artifacts were of a local nature. The fund us was modeled more grossly. The retina and choroid were represented by a cavity 800 yu.m thick covered anteriorly by a coverglass and backed posteriorly by a reflective coating of 20% reflectance in the blue, a value much higher than that in the human fundus.8 By scanning the eye without fluorescein, it was ascertained that no artifacts were introduced by scattering or reflection from the coverglasses. In vivo measurements were performed in 78 eyes from 42 individuals. The readings were performed 1 hr after an intravenous injection of 14 mg/kg of sodium fluorescein. The plasma concentration of intravenous fluorescein was obtained and used to adjust all the results to a concentration of 1 X 10"5 gm/ml. Results

Artifacts that increase VF readings. The volume in which the fluorophotometry measurements is performed is defined by the intersection of the exciting beam and the pickup path of the fiberoptic probe. This volume has a finite dimension along the optical axis. Therefore, when a sharp boundary is scanned between two compartments with different fluorescein concentrations, the photometric reading does not drop sharply at the boundary but rather tails off gradually. We will refer to the length of this transition as the depth resolution of the apparatus. (It should be noted that this phenomenon is different from the depth of field of the microscope.) Because of the linear extent of the depth

COATED ALUMINUM TRANSPARENT LUCITE

0

10

20

30 mm

Fig. 1. Diagram of model eye.

resolution, some fluorescence will be measured in the region of the compartment devoid of fluorescein that is adjacent to the compartment containing some dye. Anterior chamber contribution to vitreous reading. We scanned the model eye with different fluorescein concentrations in the anterior chamber (Fig. 2). It was seen that the tailing increased with concentration. To evaluate the contribution of the anterior chamber fluorescence to the vitreous readings in clinical situations, we measured the fluorescein concentration in the anterior chamber in vivo. In 72 eyes of 38 subjects, including normal subjects and patients with anteriorsegment inflammation, we obtained concentrations below 6 X 10~7 gm/ml 1 hr after injection. We can therefore conclude that the artifactual tailing caused by the presence of fluorescein in the anterior chamber reached meaningless values beyond the lens (Fig. 2). Lens contribution to vitreous reading. From scans performed in 20 eyes of subjects 25 to 50 years of age without cataract, we observed a lens reading equivalent to concentrations of fluorescein between 8 X 10~8 and 3 x 10~7 gm/ml. This reading was mainly caused by the natural fluorescence of the

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933334/ on 01/26/2017

Invest. Ophthalmol. Vis. Sci. May 1982

670 Zeimer et al.

5

10 15 DISTANCE (mm)

25

Fig. 2. Tailing effect in the vitreous. Artifactual reading in the vitreous is caused by fluorescein in the anterior chamber and choroid. Curves represent results obtained in the model eye, with readings in the anterior chamber and retina specified for each curve. Open circles represent data from a normal albino subject who had a retina-choroid reading equivalent to a concentration of 1.7 X 10~7 gm/ml.

lens, but there was also some contribution from the blue-light scatter not blocked by the filter. The tailing of this signal contributes to the vitreous reading, and because it resembles the tailing of the signal from the anterior chamber, we can use Fig. 2 to quantitate the effect. We can then estimate that 2 mm posterior to the lens the artifact is less than 6 x 1(T9 gm/ml. Fundus contribution to the vitreous reading. To study the tailing from the retina and choroid, the scan was done by placing the focal plane behind the retina and choroid and moving it anteriorly. The reading, as a function of the location of the probing volume, is depicted schematically in Fig. 4. The length of the probing volume is larger than the thickness of the retina and choroid; therefore, the two layers are not resolved and the

fluorescence of fluorescein present in them adds up to generate the peak of position II. We will refer to this peak as the reading on the retina-choroid. As mentioned, the depth resolution determines at what distance the tailing is relatively low (position I), that is, the distance at which the reading in the vitreous, due to the detection of fluorescence emitted by the retina-choroid, is sufficiently low. As seen below, next to the retinachoroid the depth resolution may exceed the size of the geometric intersection of the illumination beam and the collection optical path. The following tests were performed to evaluate the depth resolution in the back of the eye. When the scans were performed next to the back of the model eye (anteriorly to position II, Fig. 3), an increase in tailing was

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933334/ on 01/26/2017

Volume 22 Number 5

Vitreous

fluorophotometry

671

25 30

40 50 60 70 80

Normals 1 20-40yrsl

50

100 150 Lens Scattering Reading

200

250

Fig. 3. Optical density as a function of scattering. The dots and solid line represent data obtained with a scattering solution. The bracketed area represents the scattering measured in lenses of patients 20 to 40 years of age. noted (Fig. 2). This increase was caused by light scattering and reflection produced by the reflective coating that backed the cavity. The effect was verified by increasing the reflectance; when a reading of 5 X 10"7 gm/ml was measured in the retina, a 1% reading was recorded 3.5 mm from the retina for 20% reflectance and 5 mm from the retina for 40% reflectance. The scattering and reflection probably caused a halo that blurred the boundary of the beam's cross section and therefore increased the distance at which a low reading was recorded. The scattering by the fundus as a sphere was minimal. To evaluate the amount of tailing clinically, we performed scans on normal and diseased eyes. The readings on the retina-choroid 1 hr after injection, normalized to a plasma concentration of 1 X 10~5 gm/ml, were measured in the eyes of different groups of subjects. The retina-choroid reading in concentration equivalent was: normals, 5.4 ± 1.9 X 10~8 gm/ml (n = 7); macular edema, 4.5 ± 2.7 X 10~8 gm/ml (n = 16); hypertension, 4.4 ±

2.8 X 10~8 gm/ml (n = 27); diabetes without retinopathy, 5.3 ± 2.9 X 10" 8 gm/ml (n = 12). These results show that the peak reading on the retina-choroid is apparently not related to fluorescein leakage, since the subjects included many cases with marked leakage (e.g., macular edema). The maximal value obtained in all the examined groups was 15 X 10~8 gm/ml. Because the tailing of the retina signal in vivo could also depend on the reflectance of the fundus as well as on the retina-choroid reading, we chose an albino patient to maximize the effect of low pigmentation and possible increase in light scattering. The retinachoroid reading of this subject was 17 X 10~8 gm/ml. With the aid of the tracing obtained on this patient, we could evaluate the maximal artifactual tailing dropped to twice the background noise at 3.5 mm from the retina. Therefore we are confident that no significant artifact beyond this distance can be expected under the present conditions. The effect of the probe and beam size on the tailing has been reported 7 and was

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933334/ on 01/26/2017

Invest. Ophthalmol. Vis. Sci. May 1982

672 Zeimer et al.

Table II. Influence of beam width and probe size on performance of vitreous fluorophotometry*

m