Pharmacological treatment of vision impairing diseases often involves intraocular drug injection because the eye is well shielded by the blood-retina barrier from systemic access for -certain drug classes including therapeutic antibodies [1]. To reach the target tissue within the eye, drugs need to be transported through the vitreous humour, a highly viscous gel-like matter consisting mainly of collagen fibrils, hyaluronan and 99% water [2]. It is generally accepted that transport processes play a pivotal role in the pharmacokinetics of intraocular treatments, yet they have remained difficult to assess in the intact eye [3, 4]. Here, we devised a potentially translational approach that combines contrast-enhanced 1H-MRI and 19F-MRS to visualise and ascertain intravitral drug distribution kinetics at the macro-scale. A proof-of-concept study with a small molecule was carried out in isolated intact porcine eyes.1H-MRI and 19F-MRS were performed on a Bruker BioSpec 4,7T/40cm system equipped with a 1H bird-cage resonator (72 mm ID) and a concentric 19F Alderman-Grant resonator (25 mm ID). T1-weighted 1H-MRI was carried out with a RARE sequence (TE_eff/TR=13/694ms, 21 contiguous slices of 1mm thickness) whereas localised 19F-MRS was performed with PRESS (TE/TR=20/2500ms, 3x3x3 mm³ volume, 64 averages). Porcine eyes (n=6) were obtained from the local slaughterhouse and used within 6 hours post mortem. 50 uL of a solution of phosphate buffered saline (PBS), trifluoroethanol (TFE) and Gd-DTPA (2 mmol/L) (7:2:1 v/v) was injected intravitreally into the eye using a syringe attached to a 30 gauge hypodermic needle. The injection site was identified on an initial 1H-MR image and was used to place a volume of interest for subsequent 19F-MRS. Imaging and spectroscopy were performed alternatingly for at least one hour by acquiring a 1H-MRI volume followed by five repetitions of 19F-MRS.

In this proof-of-concept study, TFE was taken as a surrogate for a small “drug-like” molecule that carries fluorine as an intramolecular label to follow drug distribution quantitatively. The underlying notion is that the macroscopic distribution kinetics would lead to a decrease in the local drug concentration at the injection site until concentration equilibrium is reached throughout the vitreous humour. The small amount of Gd-DTPA in the injected volume allowed rapid identification of the injection site by T1-weighted 1H-MRI and provided a first pictorial view (Fig. 1) of distribution processes. Figure 2 depicts a time series of spectra obtained by 19F-MRS in order to quantitatively assess the distribution kinetics of TFE in vitreous humour. As forecast, the local drug concentration decreased steadily over the first hour after intravitreal injection (Fig. 3). Fick’s second law adapted to radial diffusion in a sphere provides a convenient approximation to the time-concentration relationship of the distribution kinetics observed in the eye:

$$\frac{\partial u}{\partial t} = D (\frac{\partial^{2} u}{\partial r^{2}})$$

with u = Cr (C: local concentration; r: radius, t: time; D: diffusion coefficient)

Macroscopic distribution kinetics in the vitreous humour even in the isolated eye turns out to be complex as it is governed both by diffusion and convection. With 1H-MRI we demonstrated that for the injected solution with a density of ~1.08 g/ml gravitational forces contributed substantially to the distribution kinetics. The data clearly showed that drug distribution in the vitreous humour and hence drug delivery to the target tissue is a slow process on a time scale of hours. Many drugs that are used to treat vitreoretinal diseases have a narrow therapeutic range. Hence, detailed knowledge about local drug concentration is essential to guide drug delivery strategies. Here, we demonstrated that localised 19F-MRS can be successfully used to quantitatively report intravitreal drug availability over time.Fluorine MRS shows great potential for directly monitoring distribution processes and the local concentration of fluorine-tagged drugs in the 1H-signal dominated vitreous humour. Prerequisite is fluorine tagging, yet many drug molecules genuinely contain fluorine. Since predominantly large molecule drugs such as antibodies are administered intreavitreally, direct monitoring is essential because their distribution kinetics may differ substantially from proxy markers. The propitious detection sensitivity demonstrated here suggests that fluorine labelling of lysine side chains enables such 19F-MRS based monitoring of antibody distribution.