A reevaluation of corneal endothelial permeability to fluorescein

The permeability of the corneal endothelium and its aqueous-cornea distribution ratio were reevaluated in the rabbit eye. Both parameters were determined in an individual eye by applying the dye first by iontophoresis and then by intravitreal injection, which allows the influence of fluorescein glucuronide on the fluorophotometric measurements to be excluded. The corneal endothelial permeability coefficient was 5.13 +/- 1.64 X 10(-4) cm min-1, and the aqueous-cornea distribution ratio was 0.25 +/- 0.06 (mean +/- S.D., n = 11) on the average, and the former was considerably greater than the previous results, while the latter was considerably smaller.     

Study of fluorescein glucuronide

Comparative studies of fluorescein and fluorescein glucuronide were carried out. Binding to human serum protein was studied using an Amicon MPS-3 ultrafiltration unit; it averaged 63% for fluorescein glucuronide and 85% for fluorescein. Intracameral penetration of both compounds was studied in the human eye, and the concentration changes of both compounds in the plasma ultrafiltrate and in the anterior chamber were analyzed, based on Davson's equation. The coefficient of entry into the anterior chamber (ki) was 0.018 +/- 0.007 h-1 (mean +/- SD, n = 10) for fluorescein glucuronide and 0.054 +/- 0.033 h-1 for fluorescein, and the former was significantly lower than the latter (P less than 0.005). The rate of loss from the vitreous (kv) was studied by injecting each compound into the vitreous of the pigmented rabbit and following the fluorescein intensity changes in it. It was 0.042 +/- 0.008 h-1 (mean +/- SD, n = 8) for fluorescein glucuronide and 0.17 +/- 0.01 h-1 for fluorescein, and the former was significantly smaller than the latter (P less than 0.001). Intraperitoneal injection of probenecid significantly decreased the kv of fluorescein but had little effection that of fluorescein glucuronide. It was suggested that fluorescein glucuronide is lost from the vitreous mainly by a passive mechanism.     

Movement of fluorescein and fluorescein glucuronide across the isolated rabbit iris-ciliary body

Movement of fluorescein and fluorescein glucuronide, a fluorescent metabolite of fluorescein, across the isolated iris-ciliary body of the albino rabbit was determined under short-circuit conditions using a modified Ussing's chamber. The permeabilities of this tissue to these dyes were calculated. The outward permeability (from the aqueous to the stromal side) of the iris-ciliary body preparation averaged 6.63 +/- 0.86 for fluorescein and 1.51 +/- 0.47 X 10(-6) cm/sec for fluorescein glucuronide, and the inward permeability (from the stromal to the aqueous side) was 1.68 +/- 0.41 for fluorescein and 1.37 +/- 0.77 X 10(-6) cm/sec for fluorescein glucuronide, respectively. Application of probenecid or ouabain decreased the outward permeability of fluorescein, but it had no significant effect on the fluorescein glucuronide movement. Application of 10(-5) M 2,4-dinitrophenol showed no significant effect on the fluorescein or fluorescein glucuronide movement, but application of 5 X 10(-4) M 2,4-dinitrophenol decreased the outward fluorescein transfer, which was also markedly suppressed by incubation at 0 degrees C. It is possible that an active transport mechanism is involved in the outward fluorescein movement across the iris-ciliary body, while the inward movement of fluorescein and also the fluorescein glucuronide movement across this tissue is mainly by passive diffusion.     

Effect of intracanalicular collagen implants on the absorption of topically applied sodium fluorescein.

Absorbable intracanalicular collagen implants were placed in both canaliculi of one eye of nine human volunteers. The other eye served as a control. Twenty-four hours later 2% sodium fluorescein was placed into both conjunctival sacs. Serial corneal fluorescein concentrations were measured with a scanning ocular fluorophotometer from 2 to 90 min after fluorescein administration. The mean corneal fluorescein concentration averaged over all time points was greater in 7 of 9 eyes with collagen implants compared to unimplanted controls. When the data from each subject were analyzed collectively, the mean corneal fluorescein concentration in the implanted eyes (1,218 micrograms/ml +/- SEM 83) was significantly greater (p less than 0.001) than the mean concentration in the control eyes (823 micrograms/ml +/- SEM 83). The use of absorbable intracanalicular collagen implants may increase the bioavailability of topically applied ocular solutions.     

Movement of fluorescein monoglucuronide in the rabbit cornea. Diffusion in the stroma and endothelial permeability

The movement of fluorescein monoglucuronide, a fluorescent metabolite of fluorescein, was studied in the rabbit cornea in vitro and in vivo. A stromal strip was exposed to fluorescein monoglucuronide, and the diffusion rate and the distribution in the stroma were measured every hr for 24 hr. The diffusion coefficient was 0.94 +/- 0.11 (+/- S.D.) X 10(-6) cm2/sec, and the saline/stroma distribution ratio was in a range of 0.67 to 0.69. The concentration of fluorescein monoglucuronide in the anterior chamber and the cornea was measured every hr for 8 hr following intravenous administration. The endothelial permeability was 4.7 +/- 1.0 X 10(-4) cm/min, and the aqueous/cornea distribution ratio was 0.56 +/- 0.05. It appears that the corneal endothelial permeability in the living eye determined hitherto from systemic administration of fluorescein is most likely the permeability to fluorescein monoglucuronide.     

Measurement of fluorescein and fluorescein monoglucuronide in the living human eye

Fluorescein monoglucuronide is a fluorescent metabolite of fluorescein, and is 1/3 to 1/34 as fluorescent as fluorescein, depending on the wavelength of excitation. After systemic administration, fluorescein glucuronide reaches concentrations many times greater than fluorescein. In order to study the effect of fluorescein glucuronide on the measurement of ocular dynamics, we devised a technique to measure fluorescein and fluorescein glucuronide in the anterior segment of the living human eye. Concentrations of each fluorophore were determined by differential spectrofluorophotometry from measurements at excitation wavelengths of 457.9 nm and 488.0 nm. Measurements were made on normal volunteers after oral and intravenous administration of fluorescein. Fluorescein was the dominant fluorophore during the first hour, while fluorescein glucuronide became dominant after 3 hours. By 6 hours there was 10 to 30 times more fluorescein glucuronide than fluorescein in the anterior chamber after oral administration, and three to ten times more after intravenous administration. The blood aqueous diffusion coefficient kd estimated from the apparent concentration of fluorescein measured at 457.9 nm was consistently greater than kd estimated from measurements at 488.0 nm. Estimates of kd, which were made on the basis of concentrations of fluorescein determined from measurements at both wavelengths, were lower than estimates based on measurements at either wavelength. These results indicate that wavelength of excitation may influence the determination of ocular parameters when systemic fluorescein is used. Care must be taken in the interpretation of measurements when metabolites of a fluorophore can interfere with measurement of the fluorophore itself.     

The blood-retinal barrier permeability to fluorescein in normal subjects and in juvenile diabetics without retinopathy

The blood-retinal barrier permeability to fluorescein was determined in 20 eyes from 17 normal volunteers (mean age 31 years) and in 20 eyes from 19 juvenile diabetics without apparent retinopathy (mean age 35 years - mean duration of diabetes 6 years). The permeability was in normal subjects (1.1 +/- 0.4) X 10(-7) cm/sec (mean +/- 2 X SD) and in juvenile diabetics (1.1 +/- 0.7) X 10(-7) cm/sec (mean +/- 2 X SD). Thus a break-down of the blood-retinal barrier cannot be demonstrated as a very early and general phenomenon in the early course of the diabetic disease. The fluorescein diffusion coefficient in the vitreous body was determined and juvenile diabetics without apparent retinopathy showed a diffusion coefficient of (0.80 +/- 0.25) X 10(-5) cm2/sec (mean +/- 2 X SD), which was the same as in normals where the diffusion coefficient was (0.69 +/- 0.46) X 10(-5) cm2/sec (mean +/- 2 X SD).     

Variations in human corneal endothelial cell morphology and permeability to fluorescein with age

Fluorophotometry with topically applied fluorescein and endothelial cell photography were performed on 80 normal subjects (age 5-79 yr). Variations in endothelial cell morphology and function, flow of aqueous humor, and intraocular pressure were recorded. The mean endothelial cell size was 332.3 +/- 46.3 micron 2. A 28% increase in endothelial cell size was measured over the eight decades (r = 0.53, P less than 0.001). The coefficient of variation of cell size also increased with age (r = 0.41, P less than 0.001). The percentage of hexagonal endothelial cells decreased by 14% (r = -0.48, P less than 0.001), while the percentage of pentagonal and heptagonal cells increased by 50% (r = 0.44, P less than 0.001) and 40% (r = 0.33, P less than 0.002), respectively, with age. The mean endothelial permeability to fluorescein was 4.03 +/- 0.63 x 10(-4) cm min-1. A 23% increase in endothelial permeability with age was observed (r = 0.44, P less than 0.001). No change in central corneal thickness or endothelial pump rate was found. Flow of aqueous humor remained stable with age, despite a 25% increase in intraocular pressure (r = 0.50, P less than 0.001). Polarization of fluorescence of fluorescein in the corneal stroma decreased with age (r = -0.46, P less than 0.001). We conclude that with age the human corneal endothelium becomes morphologically less regular and may become more permeable to fluorescein.     

Analysis of transport of fluorescein across the isolated retinal pigment epithelium-choroid using an ussing type chamber

Retinal pigment epithelium (RPE)-choroid preparations from albino rabbits were sealed in an Ussing type chamber under stabilized conditions for 3 hours. The transepithelial potential was 1.2 +/- 0.08 mV and the transepithelial resistance was 175.2 +/- 9.1 omega.cm2 (mean +/- SE, n = 16). The transport of fluorescein across the isolated rabbit RPE-choroid was studied under short circuit condition and outward (vitreous----choroid) and inward (choroid----vitreous) permeability to fluorescein were determined. The outward permeability was 1.63 +/- 0.20 x 10(-5) cm/sec and inward permeability was 0.44 +/- 0.13 x 10(-5) cm/sec (mean +/- SE, n = 8). The former was 4 times greater than the latter (p less than 0.01). The outward permeability was decreased to 1.02 +/- 0.08 x 10(-5) cm/sec (n = 7), 0.75 +/- 0.11 x 10(-5) cm/sec (n = 5), 0.67 +/- 0.11 x 10(-5) cm/sec (n = 6) by 10(-6) M of ouabain, 10(-5) M of 2,4-dinitrophenol and 10(-4) M of probenecid, respectively. Low temperatures (0.5-1.0 degree C) markedly decreased the outward permeability to 0.05 +/- 0.04 x 10(-5) cm/sec (n = 4, mean +/- SE). These results suggest that active transport plays a role in the outward movement of fluorescein across the rabbit RPE-choroid.     

The function of corneal endothelium after penetrating keratoplasty as measured with fluorophotometry]

The aim of the study was to calculate the corneal endothelial permeability (Pac) 12 and 18 months after penetrating keratoplasty in patients (aged 42-60 years) with very good prognosis for graft clarity (10 eyes with pseudophakic and 4 eyes with aphakic bullous keratopathy; 6 eyes with keratoconus; 2 eyes with granular dystrophy; 6 eyes with central inactive scars; 1 eye with early central Fuchs' dystrophy). The normal eyes (10 eyes) served as control group in persons aged 40-65 years. Each operated eye was submitted to fluorophotometry of the anterior segment with measurement of corneal endothelial permeability (Fluorotron Master, Coherent) 12 and 18 months after the surgery. The cornea thickness measurement and endothelial cell counting were performed by specular microscopy with pachymeter 12 and 18 months after penetrating keratoplasty in cases with very good prognosis for graft clarity. The mean values of Pac: 4.48 x 10(-4) +/- 1.24 cm/min after 12 months were significantly higher (p < 0.05) than in the control group (3.61 x 10(-4) + 0.51 cm/min). Neither significant changes in corneal thickness nor endothelial cell density were noted as a result of surgery. The calculation of Pac in a late period after penetrating keratoplasty revealed a stable partial exhaustion of the corneal endothelium function in cases with good prognosis for graft clarity.     

Effect of oxidized glutathione on the barrier function of the corneal endothelium

Effects of glutathione on the corneal endothelium were reexamined. Four kinds of solutions were made: oxidized glutathione (GSSG) was added to a basic solution which does not contain glutathione (GSSG-0) at a concentration of 0.03 mM, 0.3 mM or 3 mM to make GSSG-0.03, GSSG-0.3 or GSSG-3, respectively. Paired rabbit corneas were perfused separately, and the endothelial permeability (Pac) to carboxyfluorescein was determined. Between the paired corneas perfused with GSSG-0 and GSSG-0 or GSSG-0 and GSSG-0.03, there was no significant difference in the Pac. A significant difference in this factor was seen between the paired corneas perfused with GSSG-0 and GSSG-0.3 or GSSG-0 and GSSG-3 (P less than 0.01). The ratio of GSSG-0 to GSSG-0.3 for Pac, 1.18 +/- 0.16, and that of GSSG-0 to GSSG-3, 1.14 +/- 0.07, were significantly greater than the left-right ratio for Pac obtained when the paired corneas were perfused with GSSG-0, 1.01 +/- 0.10 (mean +/- SD, n = 8) (P less than 0.025). The corneal swelling rate (micron/hr) was 7.9 +/- 4.9 for the corneas perfused with GSSG-0 and 8.4 +/- 5.4 (mean +/- SD, n = 6) for those perfused with GSSG-0.3; difference was not significant. Addition of GSSG at a concentration of 0.3 mM or more to the irrigating solution was further beneficial to the corneal endothelial barrier function and a solution containing GSSG may be safer for patients with vulnerable corneas.     

Correlation of tear fluorescein clearance and Schirmer test scores with ocular irritation symptoms

OBJECTIVE: To correlate and compare the Schirmer 1 test and a new method of measuring tear fluorescein clearance with the CytoFluor II fluorometer with the severity of ocular irritation symptoms, clinical signs of meibomian gland disease, corneal fluorescein staining scores, and corneal and conjunctival sensitivity. DESIGN: Case-control study. PARTICIPANTS: Forty patients presenting with a chief complaint of ocular irritation, and 40 asymptomatic control subjects of similar age distribution. INTERVENTION: All subjects completed a symptom questionnaire, a baseline ocular examination, fluorescein clearance test (FCT), and Schirmer 1 test. MAIN OUTCOME MEASURES: The FCT was performed with a CytoFluor II fluorophotometer by measuring the fluorescein concentration in minimally stimulated tear samples collected from the inferior tear meniscus 15 minutes after instillation of 5 microl of 2% sodium fluorescein. Severity of ocular irritation was assessed with a symptom questionnaire. Schirmer 1 test, biomicroscopic meibomian gland evaluation, corneal fluorescein staining score, and corneal and conjunctival sensation scores were assessed with the Cachet-Bonnet anesthesiometer in all subjects. RESULTS: Irritation symptoms correlated with higher log tear fluorescein concentration (symptomatic 3.08 +/- 0.62 units/,microl, normal control 1.89 +/- 0.7 units/microl, P < 0.005) and lower Schirmer 1 test scores (symptomatic 12.6 mm, normal control 22.3 mm, P < 0.005). The FCT showed greater predictive value for identifying ocular irritation than the Schirmer 1 test. A fluorescein concentration of 274 units//microl eliminated 80% of the normal subjects (specificity) and identified 85% of the abnormal subjects (sensitivity). Log of tear fluorescein concentration and the Schirmer 1 test correlated with meibomian gland orifice metaplasia (2.81 +/- 0.78 units/microl and 14.47 +/- 9.53 mm in those with metaplasia vs. 1.83 +/- 0.71 units/microl and 23.14 +/- 7.67 mm in those without metaplasia, P < 0.001) and with the percentage of acinar dropout. Both log of tear fluorescein concentration and the Schirmer 1 test correlated with corneal fluorescein staining (Pearson correlation of 0.394 P < 0.0001 for Schirmer 1 test and 0.312 P < 0.005 for log of tear fluorescein). In addition, log of tear fluorescein and Schirmer 1 test scores correlated with corneal and conjunctival sensation scores (Spearman's rho for corneal sensation: log of tear fluorescein -0.38, P < 0.003, Schirmer 1 test -0.39, P < 0.002, and for conjunctival sensation: log of tear fluorescein -0.391, P < 0.001, Schirmer 1 test -0.23, P < 0.061). CONCLUSIONS: The FCT shows a greater predictive value for ocular irritation than the Schirmer 1 test. It correlates better with age, meibomian gland dysfunction, and decreased corneal and conjunctival sensation. Decreased tear clearance was identified as a risk factor for ocular irritation, even in subjects with normal Schirmer scores. This simple technique may provide new clues into the mechanism and therapy of ocular irritation.     

Acute effects of topical phenylephrine on aqueous humor dynamics and corneal endothelial permeability in man

Acute effects of topical phenylephrine, an alpha-1 adrenergic agonist, on the aqueous humor dynamics and corneal endothelial permeability were studied by means of the oral fluorescein method in 11 normal young volunteers. Twenty microliters of phenylephrine HCL (10%) was instilled in one eye and the placebo in the other eye in a double masked manner. The instillations were carried out 0.5 hour before, and 2, 4 and 6 hours after the fluorescein ingestion. Fluorophotometric measurements were carried out in the central cornea, anterior chamber and plasma ultrafiltrate, and the aqueous-cornea transfer coefficient in reference to the corneal volume (kc.ac), the transfer coefficient in the anterior chamber by diffusion (kd.pa) and by flow (kfa) were calculated in each eye. The thickness of the cornea (CT), intraocular pressure (IOP) and anterior chamber volume (Va) were also measured. No significant difference was found in the CT, kc.ac, corneal endothelial permeability (kc.ac x the thickness of the stroma), Va, IOP and kfa between the experimental and control eyes, while the kd.pa was significantly smaller in the experimental eyes (paired t-test, P less than 0.01). The iris permeability factor (kd.pa x Va) decreased significantly to 0.74 +/- 0.26 (Mean +/- SD) of the control (P less than 0.01). The aqueous flow rate (kfa x Va) averaged 0.95 +/- 0.21 of the control, and the difference from unity was not significant (P greater than 0.1).     

Blood-retinal barrier permeability to carboxyfluorescein and fluorescein in monkeys

Lipid solubility is a major determinant of permeability across the blood-brain barrier, to which the blood-retinal barrier (BRB) has many similarities. Carboxyfluorescein is a dye with about 1/1000 the lipid solubility of fluorescein, but their molecular sizes and spectral characteristics are similar. We studied the importance of lipid solubility in BRB permeability by comparing the BRB permeabilities to these two dyes. Dye in the vitreous and plasma of four monkeys was measured by fluorophotometry. The estimated inward permeability coefficients (Pin) were 11 +/- 7.4 X 10(-6) cm/min (mean and SD) for carboxyfluorescein and 21 +/- 5.9 X 10(-6) cm/min for fluorescein. The ratio of the means was 1/1.9, far from the expected 1/1000. This finding suggests that the BRB does not function as a continuous lipid membrane and that other factors are more important determinants of permeability for these dyes than lipid solubility.     

Effect of 2% fluorescein on Scheimpflug central corneal thickness measurements

AIM: To assess central corneal thickness (CCT) changes measured with Scheimpflug device following instillation of 2% fluorescein in normal subjects. METHODS: This was a prospective randomized study of 60 hospital volunteers. After baseline CCT measurements of both eyes of 40 subjects were obtained using Scheimpflug system, a drop of preservative-free 2% fluorescein, was instilled in one eye and in other eye, one drop of normal saline (control). Measurements were repeated after 1, 2, 5, 10, 20, 30, 40, 50 and 60min (continuous assessment group). Twenty subjects had baseline CCT taken, then fluorescein was instilled in one eye and measurements were taken at 1min. Ten eyes had saline rinse after 1min and 10 other eyes did not, measurements were repeated at 2min (eye rinse group). RESULTS: The mean baseline CCT for continuous assessment group was 546.2 ±32.1 μm (range, 489.0-606.0), control eyes was 546.6±30.7 μm (range, 489.0-602.0). At 1min after fluorescein instillation, CCT significantly increased by 37.0±34.0 μm (P<0.001), then decreased gradually, reaching baseline at 60min. CCT variations were not significant in control group (P>0.05). For eye rinse group, CCT mean differences between baseline and 2min were 18.2 μm (95 % CI: -54.7 to 18.3) with rinse and 26.5 μm (95% CI: -62.9 to 9.9) without rinse; paired sample tests were not significant (P>0.05). CONCLUSION: The presence of fluorescein increased CCT value to a clinically relevant level of 6.8%. Eye rinse did not significantly reduce the effect at 2min post fluorescein timepoint.     

The permeability of the corneal endothelium to fluorescein in the normal human eye

Endothelial permeability to fluorescein was measured as part of fluorophotometric studies of aqueous humor flow in 112 normal subjects whose ages span six decades. The permeability was 2.4 +/- 0.2 X 10(-4) cm/min (mean +/- SD). Females had a slightly higher permeability than males. No correlation between age and permeability was noted. The precision of this method of measurement was estimated to be +/- 22%.     

Role of fluorescein glucuronide and its metabolism in vitreous fluorophotometry

Rabbits were given fluorescein or fluorescein glucuronide intravenously. Fluorescein and fluorescein glucuronide concentrations in plasma and vitreous samples were measured by high-performance liquid chromatography. Vitreous fluorophotometry was performed using the Fluorotron Master to compare scans after administration of fluorescein and fluorescein glucuronide, and for comparison of in vivo fluorescence with in vitro high-performance liquid chromatography analysis. Fluorescein glucuronide was shown to enter the vitreous as early as 1 hr after injection. Fluorescein glucuronide was the dominant molecule in both vitreous and plasma of all rabbits at 6 hr. Because fluorescein glucuronide has a lower fluorescence than fluorescein, the fluorophotometer overestimates the vitreous concentration of fluorescein after its administration. Since fluorescein is metabolized rapidly to fluorescein glucuronide in man, entry of fluorescein glucuronide into the eye should be considered in measurements of blood-ocular barrier permeability by vitreous fluorophotometry.     

Three conventional eye drops versus a single lyophilisate. A comparative bioavailability study

PURPOSE: The ocular bioavailability of a single application of a triple dose of sodium fluorescein to the human anterior segment of the eye was studied as a novel drug delivery device. METHODS: The lyophilisate contained a fluorescein dose of 204 microg corresponding to three conventional, preservative-free eye drops of 40 microl Fluorescein SE Thilo 0.17% (68microg each) (Alcon). A single lyophilisate was applied to one eye of 22 healthy volunteers (+1 min) and three conventional eye drops (+1, 16, 31 min) were applied to their fellow eye. In this randomized, open label study, fluorophotometry was performed (Fluorotron Master IItrade mark, Ocumetrics, Mountain View, California, USA) before and +15, 30, 45, 60, 120, 180, 240, 300, 360, 420 min after application. The fluorescein concentrations of the corneal stroma (C), mid-anterior chamber (AC) were analyzed by paired t-test. RESULTS: Cornea and AC mean values (ng/ml) were significantly higher (p < 0.018, paired t-test) in the lyophilisate group up to 7 h after application, with the exception of +45 min. The mean fluorescein bioavailability from the lyophilisate was up to 11 times higher in the C and up to 8.7 times higher in the AC compared with the three preservative-free eye drops. DISCUSSION: For the first time a triple dose was delivered to the human eye with a single lyophilisate application. Significantly better bioavailability was achieved in the C and AC for up to 7 h using this new device. The treatment of glaucoma, bacterial, viral, and fungal infections, as well as dry eye syndrome, for example, will be improved using lyophilisate.     

Differential spectrofluorometry in the human vitreous: blood-retina barrier permeability to fluorescein and fluorescein glucuronide

A method is described for the separate quantitation of fluorescein and fluorescein glucuronide in the vitreous by differential spectrofluorometry. An ocular fluorometer was equppied with monochromatic laser excitation at two rapidly interchangeable wavelengths. The data analysis accounts for absorption of light in the cornea, lens, and extrinsic ocular fluorophores. Examination of seven patients with insulin-dependent diabetes and different degrees of diabetic retinopathy demonstrated that both fluorescein and fluorescein glucuronide enter the eye through the blood-retina barrier. The mean ratio between the permeabilities of fluorescein glucuronide and fluorescein was 0.9 (range, 0.3–1.9). Thus, differences in the molecular size and lipid solubility of the two substances appear to be of little or no importance for their inward penetration of the barrier. No association was found between the relative permeability and the degree of retinopathy.This study was supported by Danish Medical Research Council grant 12-8301 Offprint requests to: M. Larsen     

Carboxyfluorescein. A dye for evaluating the corneal endothelial barrier function in vivo

The relationship of the cornea-aqueous distribution ratio (r ca) and concentration in vitro was established for fluorescein and carboxyfluorescein. The value of rca for fluorescein was found to fall from 3.20 +/- 0.25 (mean +/- S.D., n V 6) to 1.78 as the concentration of the free fluorescein in the bathing medium rose from 5.8 X 10(-8) to 5.9 X 10(-5) g ml-1. For carboxyfluorescein, it remained unchanged over the same concentration range, and the average for total determinations was 1.29 +/- 0.16 (n = 20). The value of rca for carboxyfluorescein determined in vivo was 1.62 +/- 0.23 (mean +/- S.D., n = 6) and the corneal endothelial permeability to carboxyfluorescein in normal rabbits was 3.31 +/- 0.66 X 10(-4) cm min-1 (n = 11), which was 35% lower than that for fluorescein. Because of its lower endothelial permeability and a value of rca which is unchanged over a wide range of concentration, carboxyfluorescein may be better suited for the in vivo evaluation of the barrier function of the corneal endothelium than fluorescein.