A novel in vivo rabbit model that mimics human dosing to determine the distribution of antibiotics in ocular tissues.

PURPOSE: The aim of this study was to establish a novel method to predict the human ocular penetration and distribution of topical antibiotics by using a controlled rabbit model that mimics the human eye with manual blinking and tear flow. METHODS: After anesthetizing the rabbits, a single dose of commercial antibiotic formulations was given with precision directly onto the cornea. This was followed by a 30-min controlled period applying manual blinking (4 blinks/min) and a supplementary tear flow (2 microL/min) that mimics the human eye. Tear samples were collected every 5 min and after euthanasia, conjunctival, aqueous humor, iris-ciliary body, and scleral samples were collected. The corneas were mounted in perfusion chambers to determine the level and continuing rate of release of the antibiotics, the levels of which were all determined using high-performance liquid chromatography analysis. RESULTS: U.S. formulations achieved conjunctival and corneal levels (mug/g) as follows: moxifloxacin, 6.6 +/- 0.3 and 50 +/- 5; tobramycin, 3.1 +/- 1.4 and 20 +/- 5; gentamicin, <2 and <2; levofloxacin, 1.5 +/- 0.3 and 19 +/- 2; gatifloxacin, 0.9 +/- 0.1 and 11 +/- 1; and trimethoprim, <0.1 and 2 +/- 1. Japan formulations achieved conjunctival and corneal levels as follows: levofloxacin 2.1 +/- 0.8 and 12 +/- 2; gatifloxacin, 2.2 +/- 0.9 and 7 +/- 1; ofloxacin, 1.6 +/- 0.5 and 7 +/- 1; and tosufloxacin, 0.7 +/- 0.1 and 1.5 +/- 0.3 (mean +/- standard error, n = 4). CONCLUSIONS: Moxifloxacin achieved the highest levels of antibiotic in ocular tissues. In the conjunctiva and cornea, the moxifloxacin level was 3-30 times the level of other fluoroquinolones, at least twice the level of the aminoglycosides, and 25 times the level of the antibacterial trimethoprim.