The present study demonstrated clear penetration and distribution of eye drops into the vitreous cavity using a novel approach with super-long TE (i.e., UH-T2W) and 3D-real IR sequences. To the best of our knowledge, this is the first study to clearly detect the penetration of eye drugs into the vitreous cavity using standard clinical MRI. In enucleated pig eyes with topically administered Gd and 17O, signal changes could be observed visually in the vitreous cavity as well as in the anterior chamber. In contrast, no signal changes were observed visually with clinical ophthalmic solutions and saline; there were similar temporal changes in T1 and T2 values compared to those of the “No drop.”

Previous studies on the topical administration (eye drops) of Gd or 17O have observed signal changes only in the anterior chamber and not in the vitreous cavity [5,6,7]. The present study provided a visualization of drug penetration and distribution to the vitreous cavity by imaging over a longer time than previous studies and using high-sensitivity sequences. Obata et al. used D2O to visualize water distribution in the vitreous in rabbits [4]. However, D2O imaging requires 2H-MRI, and most clinical devices are only compatible with 1H; thus, its versatility may be problematic. In this study, the distribution in the vitreous was visualized via eye drop with a 1H-MRI. These findings will greatly advance future studies on intraocular aqueous flow dynamics.

We attempted, for the first time, to examine intraocular distribution by administering clinical ophthalmic solutions that had not been examined in previous studies. The use of UH-T2W was expected to allow visualization of the intraocular distribution because the T2 values of the selected solutions from the preliminary tests were sufficiently short (e.g., 295 ms for TRAVATANZ) compared to that of the vitreous cavity (approximately 1500 ms before administration). Unfortunately, the distribution could not be visualized using ophthalmic solutions. The T2 value of the original 17O solution was 50 ms, and that of Gd was not measurable (quite short), which would have allowed visualization of the intraocular distributions because of the overwhelmingly short T2 value. Applying an even longer TE than that used for UH-T2W, visualization might be feasible when using clinical ophthalmic solutions. Nevertheless, given that signal changes even in the anterior chamber were not observed in the present study, ophthalmic solutions are considered insufficiently effective with respect to altering the T2 values for the vitreous cavity, compared to Gd and 17O. For other viewpoints, there may be less penetration of ophthalmic solution into the vitreous cavity. A previous study demonstrated that azone, a commonly used penetration enhancer, can increase the signal in the anterior chamber [5]. A combination of ophthalmic solutions and penetration enhancers or viscosity-increasing agents may allow visualization of distributions in the vitreous cavity.

In this study, 3D-real IR for T1 changes and 3D-T2W with super-long TE for T2 changes were found to be useful for visualizing intraocular distribution. Both sequences are 3D acquisitions, which facilitate the evaluation of detailed structures, such as distribution pathways to the vitreous. However, 3D sequences require long acquisition times, and can cause motion artifacts attributable to eye movements or blinking during scanning, which will inevitably pose problems for future human studies. To address these concerns, Tomiyasu et al. employed a single-shot sequences (i.e., HASTE) in their study on humans [7]. We therefore propose a new approach based on the application of a super-long TE using the HASTE technique, which could provide high sensitivity to T2 changes imaging, comparable to that obtained using 3D-T2W with a TE of 3200 ms (Fig. 4). A further concern is the long data duration of the 3D-T2W, which may cause blurring artifacts. Sample images obtained using 3D-T2W and HASTE sequences are shown in Fig. 6. The 3D-T2W with a TE of 3200 and 4500 ms shows more pronounced blurring artifacts than images obtained using a TE of 500 ms. In contrast, the HASTE sequence provides images without blurring artifacts, even with a TE of 3000 ms. Moreover, images can be obtained within a few seconds per slice. Thus, although it is a 2D acquisition, it can provide motion-robust and high temporal resolution. With the application of UH-T2W, HASTE would be appropriate for observation of aqueous flow dynamics with high temporal resolution in the early phase, and 3D imaging would be appropriate for a detailed anatomical interpretation of intraocular distribution.

Sample images with 3D T2-weighted and HASTE sequences before topical eye drop administration. Ultra-heavily T2-weighted sequences (TE3200 and TE4500) had blurring artifacts, whereas the HASTE sequence with super long TE (HASTE3000) provided artifact-free images

Furthermore, the use of high-sensitivity sequences suggests that drop volume can be reduced. Our study showed visualization of intraocular distribution with smaller volumes than in a previous study (0.5 mL in our study, 0.92–1.37 mL in a previous study [7]), but still too large. Future studies are needed to investigate whether signal changes can be observed with a normal dosage (i.e., 1 drop = 0.05 mL) in humans. In addition, several previous studies on Gd eye drop dacryocystography in humans have administered it at a 1:100 dilution [14, 15]. In the present study, the original solution was administered, which may have resulted in significant changes in the signal. We should determine the appropriate dilution rate while preventing adverse effects for living human study.

We also established that there are differences between Gd and 17O with respect to the speed and pattern with which these drugs penetrate the vitreous cavity. Compared with Gd, we found that 17O administered via eye drops was characterized be a more rapid distribution, which is consistent with the findings of a previous study on glymphatic water transport in the rat brain in vivo [16]. In addition, the penetration of Gd into the vitreous cavity may be initiated from the peripheral through the sclera, whereas in contrast, 17O appears to be distributed from the anterior chamber side into the vitreous cavity (Fig. 5). We suspect that these differences in distribution speeds and patterns are attributable to certain pharmacodynamic properties of the drugs, such as differences in molecular weight and viscosity, which is a topic warranting further investigation.

In the present study, we carried out eye drop experiments using “enucleated” pig eyes that are inexpensive and readily available, because these were preliminary experiments and the selection of suitable solutions for human studies was also our goal. Following the administration of Gd and 17O, distribution in the anterior chamber was observed within 0.5 h, and we were able to visualize the distribution to the vitreous cavity at 1 h post-administration. In some previous studies, 17O distribution within the anterior chamber was detected after 4 min in sedated rabbits [6], and also after a few min in living human [7]. Moreover, although in both these studies, monitoring was continued for up to approximately 40 min after administration, no distribution was detected in the vitreous cavity. The findings from these in vivo studies are consistent with our ex vivo results. In the present study, however, we failed to observe any drainage of the administered solution from the vitreous cavity (images were obtained until 62 h; data not shown). This finding suggests that the temporal changes in intraocular water flow out may be considerably different from those in living humans.

Given their morphological and functional similarities to those of the human eyes, the eyes of pigs are often used as an ex vivo animal model in vision sciences and ocular pharmacology research [17,18,19,20,21]. It has been reported that the porcine sclera has almost the same permeability as that of the human sclera [22]. Moreover, the structure of the cornea, vitreous properties, and retinal morphology do not differ significantly from those of human eyes. In addition, the diameter of the cornea and the size of the eyeball are similar to those in humans. The major difference between these two mammals lies in the thickness of the cornea, which is approximately 1.7 times thicker in pigs (877 ± 14 μm) [22] than in humans (521 ± 32 μm) [23]. This may result in poorer drug permeability of the porcine cornea than in the human cornea. Accordingly, this can be considered a limitation of the present study. Consequently, as a next step, we will proceed to investigate intraocular distribution via the administration of Gd or 17O eye drops using high-sensitivity sequences in human studies.