Measurement of Blood-Retinal Barrier Permeability A Reproducibility Study in Normal Eyes

Transcription

1 Measurement of Blood-Retinal Barrier Permeability A Reproducibility Study in Normal Eyes Prirpal 5. Chahal, Philip J. Chowienczyk, and Eva M. Kohner Fluorescein penetration into the posterior vitreous depends on plasma-free fluorescein concentration and blood-retinal barrier (BRB) permeability. The reproducibility of two methods of deriving BRB permeability was studied in 19 normal eyes of 14 subjects using vitreous fluorophotometry on two separate occasions. Plasma-free fluorescence was measured at intervals over 1 hr and posterior vitreous fluorescence was measured before (background scan), within 6 min (bolus) and at 60 min (measurement) after intravenous fluorescein (14 mg kg"'). A computer algorithm subtracted background fluorescence from the measurement scan which was then corrected for signal spread by using a "spread" function derived from the bolus scan. BRB permeability coefficient and vitreous diffusion coefficients were derived by fitting a mathematical model to the plasma and corrected vitreous fluorescence data. A permeability index was also calculated by dividing the area under the vitreous fluorescence by the area under the plasma fluorescence curve. There were no significant differences in the results between right and left eyes. Mean ±SD values on first and second occasions for all eyes were permeability coefficient: (1.91 ± 034) and (2.08 ± 0.95) X 10" 7 cms 1 ; diffusion coefficient: (1.33 ± 0.68) and (1.19 ± 0.54) X 10 s cm 2 s~'; and permeability index: (2.05 ± 1.03) and (2.11 ± 1.02) X 10~ 7 cm-s" 1. Mean variability (%) ± SD for permeability coefficient, diffusion coefficient, and permeability index was 20.6 ± 11.3, 22.5 ± 19.5, and 22.0 ± 15.8, respectively. Correlation between permeability coefficient and permeability index was r = 0.96 (P < 0.001) on the first and r = 0.97 (P < 0.001) on the second occasions. Permeability coefficient and permeability index are comparable and reproducible measurements of BRB permeability. Invest Ophthalmol Vis Sci 26: , 1985 The measurement of posterior vitreous fluorescence with fluorophotometry after fluorescein administration is an indirect assessment of blood-retinal barrier (BRB) function. 1 The penetration of fluorescein and its subsequent diffusion is dependent on the plasmafree fluorescein concentration, BRB permeability, and diffusion of fluorescein in the vitreous body. Recently, there has been an emphasis on deriving BRB permeability. 2 ' 3 However no reproducibility studies have been done to assess the variation in the measurement of permeability in the same eye. This is most important if vitreous fluorophotometry is to be used to compare or follow quantitative changes in BRB function particularly as in prospective studies a normal eye may become abnormal or diseased eyes may From the Department of Medicine, Royal Postgraduate Medical School, Hammersmith Hospital, Du Cane Road, London. Presented at the ARVO meeting, Sarasota, Florida, May Supported by the British Diabetic Association, the Juvenile Diabetes Foundation and by the University of London Central Research Fund. Submitted for publication: June 7, Reprint requests: Dr. P. Chahal, Royal Postgraduate Medical School, Du Cane Road, London W12 OHS. change in severity. We report the results of a study in normal eyes comparing the reproducibility of two methods of deriving BRB permeability using a new commercial vitreous fluorophotometer (Fluorotron Master, Coherent Radiation; Palo Alto, CA). Subjects and Procedure Materials and Methods The study had been approved by the ethics committee and informed consent was obtained from all subjects. One or both eyes (n = 19) of 14 normotensive healthy volunteers with a mean age of 35 yr (range, yr) were tested by the same observer on two separate occasions with a mean interval of 5 wk (range 1-15). All subjects had normal renal and hepatic function. Tropicamide (1%) was used to obtain pupillary dilatation and paralysis of accommodation. Background fluorescence was measured before the intravenous bolus injection of 14 mg- kg" 1 of 20% sodium fluorescein and within 6 min (bolus scan) and 60 min (measurement scan) after injection. Ampoules of sodium fluorescein with the same batch number were used on both occasions in each subject. 977

2 978 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / July 1985 Vol. 26 Blood was sampled from a peripheral arm vein for plasma fluorescence estimations at 2, 5, 10, 15, 30, 45, and 60 min after fluorescein injection. Measurement of Plasma Fluorescence Fifty microliters of supernatant plasma was diluted in 10 ml phosphate buffer (ph 7.4 at 22 C) and the total plasma fluorescence in units of fluorescein (/ug-mp 1 ) was measured with the fluorophotometer. The remaining plasma was placed into microfiltration chambers with anisotropic and hydrophilic YMT (Amicon) membrane filters. A protein-free ultrafiltrate was obtained by centrifugation (1500Xg) at C. Fluorescence in the ultrafiltrate was measured after diluting 50 /*1 of the ultrafiltrate in 10 ml of phosphate buffer. Our studies (submitted) have shown that the ultrafiltrate contains unbound (free) fluorescein and its weakly fluorescent metabolite, fluorescein glucuronide, in varying proportions. Consequently, the term "plasma free fluorescence" is preferred for ultrafiltrate measurements of fluorescence and the term "plasma free fluorescein" will be used for theoretic considerations. Vitreous Fluorophotometry Technical details and performance of the vitreous fluorophotometer have been described elsewhere 4 and will only be summarized here. The optical scanning system is linked to the eye such that no contact lenses are required. The eye is aligned using a split-image range finder while the subject fixates on a target that remains visible during scanning. A computer controls the movement of the optical system which measures fluorescence at 0.25-mm steps starting from an apparent point behind the retina to end in front of the cornea. Fluorescence in equivalent units of fluorescein (ng'mr 1 ) vs distance from the retina is displayed graphically on a visual display unit and stored onto magnetic diskettes. The axial resolution of the instrument has been reported to be 1.5 mm. This implies that measurement of fluorescence is unreliable at a distance closer than 1.5 mm to an area of high fluorescence such as at the choroid-retina because of signal spread. The signal spread can be expressed as a function of the intensity of adjacent fluorescence and can be derived from the bolus scan. A computer algorithm subtracted the background fluorescence from the bolus and measurement scans and then further corrected the measurement data for signal spread. 5 In practice this correction mostly influenced fluorescence data within 2.0 mm of the choroid-retinal peak. This corrected vitreous fluorescence data and the plasma ultrafiltrate fluorescence were analyzed by computer. Calculation of Permeability and Diffusion Coefficient The mathematical model assumes the following: (1) the amount of fluorescein traversing the BRB per unit time is determined by the difference in concentration between the blood and vitreous and the constant of proportionality is the permeability coefficient; (2) diffusion of fluorescein in the vitreous is described by a simple first-order diffusion equation; (3) the leakage of fluorescein and its diffusion into the vitreous occurs radially from the retina and there is no interference from backward diffusion of fluorescein from the anterior chamber; (4) the effects of any inward or outward active transport mechanisms for fluorescein are minimal during the period of measurement; and (5) fluorescent metabolites of fluorescein do not affect measurements. The concentration (C) of fluorescein at a distance (r) from the center of the vitreous and at time (t ) may be related to the concentration of plasma free fluorescein (C p ) and to the permeability coefficient (P) and diffusion coefficient (D) by the expression: where C(r, t.) = f Jo C p(t - t,)f(r, t)dt (1), ^ (2). Equation 2 is a modification of the expression derived by Larsen et al. 3 This modification allowed the analysis to be performed on a microcomputer. An itterative procedure was used to determine P and D from equation 1. The sum of the squares error between values of C(r, ti) computed from equation 1 and the experimental values of C(r, t,) from r, to r 2 was used as criterion for obtaining the best fit computer curve for the observed vitreous fluorescence scan (Fig. 1). Figure 1 shows a vitreous fluorescence scan after correction. Despite the corrective algorithm, it was considered that fluorescein concentration within 2 mm of the retina may not be reliable and r, was fixed at 2 mm from the retina. The distance from the retina r 2 was where the vitreous fluorescein concentration fell to less than 1 ng- ml" 1, the lower limit of fluorescence detection for the instrument. It is not possible to measure the radius (a) of the eye with this particular fluorophotometer and a standard value of 1.2 cm was used for all eyes. Permeability Index A simple index of permeability used in previous studies 6 involves measuring the penetration of fluorescein into the posterior vitreous by estimating the area under the vitreous fluorescence curve vs distance and

3 No. 7 BLOOD-RETINAL BARRIER PERMEABILITY / Chohol er ol. 979 dividing this by the total "exposure" to plasma-free fluorescein. Thus, at time t = t ( the permeability index can be derived from our data by: Area under posterior vitreous fluorescence curve Area under the plasma free fluorescence curve (integrated from t = 0 to t = t t ) The permeability index has the same units (cm«s~') as the permeability coefficient. Measurement of the permeability index has theoretic disadvantages. It assumes one-dimensional diffusion of fluorescein across a plane retina and it is necessary to include measurements of fluorescence near the retina where they may be unreliable despite the correction. The area under the vitreous scan from r! = 0 to r 2 (as defined above) was measured. The areas under the curves were measured using Simpson's trapezoidal rule. It was assumed that the concentration of plasma free fluorescein varied linearly from 0 ng«ml" 1 at time t = 0 to the value measured at 2 min and the area under the curve was calculated up to 60 min. Statistical Methods As the results in right and left eyes in the same subject may be correlated, these eyes were analyzed separately. A measure of reproducibility or variability can be obtained by expressing the absolute difference between the values on the first (x) and second (y) occasions as a percentage of their mean, ie, variability (%) = - y (x + y)/2 X 100. The overall percentage change, increase or decrease, between the two occasions can also be calculated from the expression change(%) = x -y (x + y)/2 X 100. The unpaired t-test was used to assess whether the differences between mean values in right and left eyes were significant. The paired t-test was used to assess the significance of a change in measurement in the same eye. A value of P < 0.05 was considered significant. Results In one subject, vitreous fluorescence measurement in one eye on the second occasion was so markedly increased from the first that doubts were raised of its validity. The subject was short-sighted (-1.25 D), but there was no satisfactory explanation for the discrepancy. In this subject, the procedure was repeated on en c r, 10 '1 '2 15 mm Fig. 1. Posterior vitreous fluorescence after correction (ng- ml" 1 ) versus distance from the retina, r, and r 2 are the boundary conditions for computer analysis and a is the radius of the eye. a third occasion and the results were more consistent with those obtained on the first occasion, and the results of the second occasion were excluded from analysis. Plasma-free Fluorescence Plasma-free fluorescence measurement and its percentage of the total plasma fluorescence vs time are shown for both occasions in Figure 2. There was no significant difference between the mean plasma-free fluorescence measurements on the two occasions at the times of sampling. The variability in measurement of plasma-free fluorescence was greater at the beginning than towards the end of the hour. In some subjects on some occasions, the plasma fluorescence peak value was delayed until after 2 min. The plasmafree fluorescence as a percentage of the total plasma fluorescence increased from 12 to 15% at the beginning to 23-26% towards the end of the hour. Permeability and Diffusion Coefficients Table 1 summarizes the results and their variability for permeability coefficient, diffusion coefficient, and permeability index for right and left eyes separately and combined. The values for permeability coefficient in right and left eyes on the two occasions is shown in Figure 3, which also shows the wide variation between individuals on both occasions. There was no significant difference in permeability coefficient or its variability between right and left eyes. The permeability coefficient tended to be higher on the second occasion with an overall increase of about 9.2% and a variability of 20%. The difference in permeability coefficient between the two occasions in the same eye

4 980 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / July 1985 Vol a - u c 0) ) o-p w 15 10!J If k It ^ '= 10 - s E 5 - I? Time (mins) Fig. 2. Concentration of plasma free fluorescence (lower figure) expressed as a percentage of the plasma total fluorescence (upper figure) versus time on first () and second (O) occasions (vertical bars = sd). was not significant (P > 0.1). In seven subjects where both eyes had been tested simultaneously in 12 instances there was no significant difference in per- Table 1. Permeabilities and diffusion coefficient on first and second occasions and their variability in 10 right (R) and 9 left (L) eyes separately and together (R&L) Permeability coefficient (cm s" 1 X I0" 7 ) Permeability index (cm s" 1 X 10" 7 ) Diffusion coefficient (cm 2 s~' X 10" 5 ) Eye R L R&L R L R&L R L R&L Mean ± sd. Range in parentheses. First 1.96 ± 1.00 ( ) 1.86 ±0.93 ( ) 1.91 ± ± 1.16 ( ) 2.14 ±0.94 ( ) 2.05 ± ±0.31 ( ) 1.66 ±0.83 ( ) 1.33 ±0.68 Second 2.10 ±0.98 ( ) 2.06 ± 0.98 ( ) 2.08 ± ± 1.06 ( ) 2.21 ± 1.02 ( ) 2.11 ± ± 0.24 ( ) 1.36 ±0.73 (O.8O-3.OO) 1.18 ± Variability (%) 17.4 ± 10.5 ( ) 24.2 ± 11.5 ( ) 20.6 ± ± 16.8 ( ) 18.9 ± 14.9 ( ) 22.0 ± ± 11.0 ( ) 30.1 ±24.4 ( ) 22.5 ± 19.5 a> E i_ a; a. 0-1 I First I Second Fig. 3. Changes in permeability coefficient in right ( ) and left (O O) eyes on first and second occasions of testing. meability coefficient between right [(2.49 ± 1.84) X 10~ 7 cm-s" 1 ] and left [(2.20 ± 1.37) X 10" 7 cm-s" 1 ] eyes. The variability of the diffusion coefficient was 22% with a wide individual variation. There was an overall decrease in the measurement of diffusion coefficient on the second occasion of 6.1%. Permeability Index The permeability index results were similar to and as reproducible as the permeability coefficient for right and left eyes separately and combined. This is not surprising as they were highly and significantly correlated (Figs. 4A, B). The relationship between permeability coefficient (P) and the permeability index (PI) is given by the combined regression equation for right and left eyes of P = 0.88 PI (r = 0.96; P < 0.001) and for the first and P = 0.90 PI (r = 0.97; P < 0.001) on the second occasion. The mean PI/P ratio for all 19 eyes was 1.08 ±0.17 (range, ) on the first occasion and 1.01 ±0.13 (range, ) on the second occasion. There was an overall percentage increase of 3.2% in permeability index on the second occasion. There were no significant differences in permeability index between the right and left eyes. Discussion In this study, the two methods of deriving BRB permeability produced comparable and reproducible

5 No. 7 BLOOD-RETINAL BARRIER PERMEABILITY / Chohol er ol. 981 measurements in normal eyes. The permeability index is easier to derive and may be a good way of comparing data from different fluorophotometers. In our study, the ratio between the permeability index and permeability coefficient at 1 hr was Larsen 2 calculated an equivalent ratio to be 1.28 using specimen plasma-free fluorescein data and a diffusion coefficient of D = 6 X 10" 6 cm 2 -s~'. Although there was wide variation between eyes of different subjects, the BRB permeability was similar in right and left eyes of the same subject. The values for permeability tended to be higher on the second occasion. It is possible that familiarity with the procedures may have resulted in better cooperation by subjects. This "learning effect" is difficult to exclude though it may be assessed by repeating the measurements on more than two occasions. Our measurements of permeability coefficient in normal eyes are a little higher than those reported elsewhere using the same fluorophotometer 5 and this is probably due to differences in methods of data analysis. The bolus method of administration resulted in greater lability of plasma free fluorescence values in the first 5 min. In some subjects on some occasions, the mixing of fluorescein in peripheral blood was incomplete 2 min after the bolus. This indicates that measurement of fluorescence in peripheral blood within the first 2 min does not necessarily reflect the concentration of fluorescence in the retinal arterial blood. Theoretically, the free plasma fluorescein fraction in the bolus blood to the eye would be high because of transient saturation of protein-binding sites by the fluorescein. The magnitude and significance of this bolus effect is uncertain. Some insight would be gained by measuring permeability in the same individual using longer infusion periods during which peripheral fluorescence would more accurately reflect the concentration reaching the retina. In this study, the bolus method was preferred so that comparisons could be made between normal eyes and other eyes in which fluorescein angiography would be integrated with vitreous fluorophotometry. Oral fluorescein is conjugated to a monoglucuronide 78 which has about 5% of the fluorescence of fluorescein. Our studies have shown that intravenous fluorescein is also rapidly glucuronidated (submitted). The increase in percentage free fluorescence observed in this and other studies 9 is due to the glucuronide metabolite being less bound to plasma protein and constituting an increasing proportion of the fluorescence in the ultrafiltrate particularly after 30 min. Whether or not the metabolite also appears in the vitreous is speculative but it probably contributes very little to the overall vitreous fluorescence measured at 1 hr in normal eyes with intact BRB. However, ~ 1 - ~ R: y =0. 85x (r=0. 98; p<0. 001) o L: y =0. 97x (r=0. 97; p<0. 001) «O 4> O o Permeability index (cm s x 10 ) R: y=0.89x (r=0.96; p<0.001) o L: y=0.94x (r=0.98; p<0.001) Permeability index (cm s x 10 ) Fig. 4. The correlation between permeability coefficient and the permeability index on the first (A) and second (B) occasions in right () and left (O) eyes. variation in production of this metabolite between subjects or in the same subject at different times explains the greater variation in the percentage free plasma fluorescence towards the latter part of the hour. There are errors in the measurement of vitreous fluorescence. A variability of 13% has been reported when measurements of fluorescence at discrete points from the retina are repeated at 2-min intervals. 4 Differences in signal spread and in the use of the corrective algorithm may be sources of variability though the fluorescence values within 2 mm of the retina were not used for deriving permeability coefficient. Interestingly, the corrected fluorescence values

6 982 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / July 1985 Vol. 26 close to the retina were used to derive the permeability index, which still correlated well with the permeability coefficient. This vitreous fluorophotometer cannot measure the radius of the eye and the mathematical model assumed a constant value of 1.2 cm. If gross errors in refraction are excluded (±6 D) the diameter of most eyes is between cm and, therefore, a radius of 1.2 cm is a reasonable average value. 10 Differences in the radius could account for some differences between subjects though not within the same subject. The estimation of permeability index did not depend on a radius value and this may be an advantage. Estimations of blood-retinal barrier permeability will only be as reliable as the data collected from the eye and plasma. Until all artefacts can be accounted for, there will always be uncertainty in these measurements. Key words: blood-retinal barrier, fluorophotometry, fluorescence, permeability, diffusion coefficient, variability Acknowledgment The authors thank Margaret Foster for technical assistance. References 1. Cunha-Vaz J: Vitreous fluorophotometry. In The Blood-Retinal Barriers, Cunha-Vaz JG, editor. New York, Plenum Press, 1980, pp Larsen J, Lund-Andersen H, and Krogsaa B: Transient transport across the blood-retinal barrier. Bull Math Biol 45:749, Zeimer RC, Blair NP, and Cunha-Vaz JG: Pharmocokinetic interpretation of vitreous fluorophotometry. Invest Ophthalmol Vis Sci 24:1374, Zeimer RC, Blair NP, and Cunha-Vaz JG: Vitreous fluorophotometry for clinical research. 1. Description and evaluation of a new fluorophotometer. Arch Ophthalmol 101:1753, Zeimer RC, Blair NP, and Cunha-Vaz JG: Vitreous fluorophotometry for clinical research. 2. Method of data acquisition and processing. Arch Ophthalmol 101:1757, Krogsaa B, Lund-Andersen H, Mehlsen J, Sestoft L, and Larsen J: The blood-retinal barrier permeability in diabetic patients. Acta Ophthalmol 59:689, Chen SC, Nakamura H, and Tamura Z: Studies on the metabolites of fluorescein in rabbit and human urine. Chem Pharmacol Bull (Tokyo) 28:1403, Chen SC, Nakamura H, and Tamura Z: Determination of fluorescein and fluorescein monoglucuronide excreted in urine. Chem Pharmacol Bull (Tokyo) 28:2812, Lund-Andersen H, Krogsaa B, and Jensen PK: Fluorescein in human plasma in vivo. Acta Ophthalmol 60:709, Abrams D: Variations in components of the optical system. In Duke-Elder's Practice of Refraction, 9th ed. Revised by Abrams D, London and Edinburgh, Churchill-Livingstone, 1978, p. 33.