One of the critical issues of topically administered ophthalmic drugs is that their efficacy is limited by the fast drug clearance due to pre-corneal fluid drainage; consequently, frequent administrations are required. Therefore, various drug delivery systems (DDS) have received increased attention to enhance the efficacy of drugs on the corneal surface.
Aside from this, many limitations make it hard to deliver drugs aimed to treat eye posterior segment diseases, such as diabetic retinopathy and age-related macular degeneration (AMD). In fact, topical ocular medications do not reach the back of the eye; moreover, systemic administration is rarely used because of the small volume of the eye and the presence of the blood retinal barrier (BRB) [1
A lot of research is currently being done to improve transscleral delivery, which might offer the advantage of removing the need to breach the walls of the eye; many transscleral delivery systems, also associated to iontophoresis, are therefore at different stages of development. However, to date, the majority of treatments of the posterior segment, such as retinal and choroidal disorders, require the intravitreal pathway. Intravitreal injection (IVI) is currently considered the most validated option—although it is invasive and associated with serious side effects—for the delivery of large molecules such as anti-vascular endothelial growth factors (anti-VEGF antibodies), whose use has reached an exponential growth in recent years due to the progressive expansion of their clinical applications [2
]. Nevertheless, the periocular pathway, including the retrobulbar, peribulbar, subtenon and subconjunctival routes, and even topical delivery, continue to be explored.
To overcome the limitations of conventional eye drops in corneal/conjunctival administration, and of invasive injection in intraocular administrations, or of surgery-implanted cannulas in periocular administration, in the last decades, several ophthalmic formulations, such as drug-loaded hydrogels and contact lenses targeted to the anterior segment, or ocular implants and physical devices destined to the back of the eye, have been proposed.
Hydrogels are three-dimensional, cross-linked networks of either synthetic or natural water-soluble polymers with great potential in several applications, such as drug delivery, cell encapsulation and tissue engineering [3
]. Various advantages make them interesting—their aqueous environment can exert some protection towards cells and labile drugs (such as peptides, proteins, DNA and oligonucleotides) and they have a significant role in transporting nutrients to cells. As a result, they are attractive for various ophthalmic applications, among which are corrective soft contact lenses [4
], adhesives for ocular wound repair [5
], potential vitreous substitutes [6
] and drug vehicles [7
]. Regarding the latter application, to our knowledge, most hydrogels on the market are targeted to the anterior segment (such as Pilopine HS, Zirgan and Pilogel), due to their ability to increase viscosity and mucoadhesive properties [8
]. Conversely, just few hydrogels have been already approved by the FDA and EMA for intraocular injectable applications; as an example, Akten (Akorn, Buffalo Grove, IIlinois), a 3.5% lidocaine gel, was approved by the FDA for all ophthalmic procedures in October 2008, including intraocular procedures [9
Other hydrogels are currently on the market for ophthalmic application other than drug delivery purposes—as an example, ReSure Sealant is an in situ gel approved to seal clear corneal incisions following cataract surgery [11
Recently, contact lenses for drug controlled delivery to the anterior chamber have been developed, but several challenges are still arising regarding the limited drug release, the strict regulatory issues and the high cost of clinical studies [12
On the other hand, the crucial need to reach the eye posterior segment through less invasive strategies other than repeated injections has boosted the development of slow-release implants that can be placed at once at various ocular sites. Currently, intraocular implants that allow sustained drug release in the posterior segment are at different development stages [13
]. Most of these consist of non-biodegradable polymers, such as silicone, polyvinyl alcohol and ethylene vinyl acetate, from which long-lasting release of the entrapped drugs occurs [14
]. However, they require surgical intervention and need removal or replacement by new implants. On the contrary, biodegradable polymers such as poly(lactic) acid and poly(lactic-co
-glycolic) acid offer the advantage of releasing the drug at the same time that the polymer degrades in the target site, avoiding the need of surgical removal [15
To date, several ocular implants designed for the treatment of severe indications affecting the posterior segment of the eye, including macular degeneration, are on the market or are undergoing clinical trials. Among them, non-biodegradable Vitrasert, Retisert, Medidur, Iluvien and biodegradable Posurdex, Ozurdex and Surodex must be cited [17
]. Most of these implants are loaded with small active compounds, such as fluocinolone acetonide, dexamethasone and ganciclovir; meanwhile no biologics-carrying implants are available in the market, some being, however, in the pipeline [21
Anti-VEGF therapy, playing a central role in the pathogenesis of choroidal neovascularization, has revolutionized the medical management of diabetic retinopathy and of AMD [22
]. Currently, the most common anti-VEGF agents are pegaptanib (Macugen), bevacizumab (Avastin) and ranibizumab (Lucentis) [23
], followed by other emerging macromolecular drugs already in clinical trials, among these being Fovista (Ophthotech, Princeton, NJ, USA), a platelet-derived growth factor aptamer currently in phase III clinical trials [25
], and designed ankyrin repeat proteins [26
The use of intravitreal administration in anti-VEGF therapy is still presenting some problems—most drugs are rapidly cleared from the vitreous humor, inducing the need of repeated injections that can cause side effects, such as endophthalmitis, retinal detachment, hemorrhage and poor patient tolerance [27
]; other drugs induce local toxicity when administered at their effective dose, causing side effects and possible retinal lesions [28
]. For these reasons, strategies that can deliver sufficient drug concentrations to this anatomic region in a less invasive manner and with less frequent doses, such as sustained-release DDS, represent an area of active interest in the ophthalmology community. In the last decades, different technologies have been proposed to this aim, including the use of nanomedicine [29
]. Therapies based on nanotechnologies, such as lipid and polymeric nanocarriers, present several advantages, allowing a precisely targeted drug delivery and controllable release of the therapeutics [32
]; moreover, the stability and half-life in the vitreous of entrapped drugs might be enhanced, thus reducing the frequency of administration and, consequently, diminishing their toxicity [33
]. Therefore, depending on particle charge, surface properties and relative hydrophobicity, nanoparticles (NP) can be designed to be successfully used in sustained ocular therapy [34
], both in the anterior and posterior segments [35
]. Studies have shown that albumin NP can serve as a very efficient drug delivery system for retinal diseases, such as cytomegalovirus retinitis, as they are biodegradable, non-toxic and have non-antigenic properties [36
]. Moreover, NP prepared with natural polymers, such as chitosan, increased the intraocular penetration of loaded drugs, due to their ability to make contact intimately with corneal and conjunctival surfaces [37
]. In the past decades, several hydrophilic polymeric particles have been proposed as ocular DDS composed of various biodegradable polymers, such as poly(lactic acid) [38
], poly(alkyl cyanoacrylate) (PACA) [39
-glycolic acid) (PLGA) [40
] and poly(ɛ-caprolactone) (PECL) [41
]. However, one of the main barrier-hindering clinical trials of these innovative systems is the requirement to ensure the safety of nano-microsystems and of their biodegradation products in the eye [29
Another appealing approach of drug delivery to the posterior ocular segment consists in vesicular systems such as intravitreal-injectable liposomes (i.e., the ocular liposomal Verteporfin (Visudyne). They provide sustained drug delivery for weeks or even months, but up until today, most of them are only pre-clinically investigated, with few in clinical use [43
Recently, preclinical trials have centered around the interesting formulation of nanocomposites, consisting of nanoparticulate systems dispersed into a hydrogel matrix that provide an additional diffusion barrier to drug release, eliminating the burst effect and extending the release profiles of the entrapped drugs [44
In the literature, various strategies have been proposed to deliver drugs into the eye in a more controlled manner—in the first part of this review, special attention will be given to the thermosensitive approach, considering the different typologies and action mechanisms. The second part deals with the most widespread in vitro models employed to investigate the functionality of novel ophthalmic DDS.
3. In Vitro and In Silico Models to Test Intraocular DDS
Currently there is a lack of convenient platforms for the treatment of posterior eye diseases, as several aspects of ocular drug delivery, such as distribution, clearance and overcoming barriers by drugs, are not completely understood. Meanwhile, corneal and conjunctival epithelia play an important role for topically administered ophthalmic drugs in the development of alternative therapy for the back of the eye. It is important to know the crucial role of the retina, the thin light-sensitive membrane that lines the inner surface of the eye and that, together with the retinal pigment epithelium (RPE), constitutes the blood retinal barrier (BRB) [97
]. Intravitreally administered drugs distribute in the vitreous and to the surrounding ocular tissues. Their elimination from the vitreous cavity occurs via blood-ocular barriers and aqueous humor outflow.
In parallel, the knowledge of pharmacokinetic (PK) profiles of intraocular administered drugs is essential to determine the minimum dosing frequency to achieve the maximum therapeutic concentration. Several researchers analyzed the PK parameters of intravitreal-injected drugs in rabbit eyes with the employment of animal tests [98
]. Overall, traditional PK studies of eye dissection yielded important information about the mechanisms, even if these approaches are affected by some limitations. The employment of magnetic resonance imaging (MRI) was useful in assessing drug penetration through ocular barriers, the location of ocular and periocular depots, the release kinetics from ocular implants and the clearance process kinetics [99
Moreover, the use of in vivo models makes the results difficult to evaluate, to compare and to correlate, especially whenever eye dissection is required [100
]. The challenge is to design alternative experimentally controlled models that give more reproducible data, reducing animal use, and with consequent cost reductions and ethical advantages.
The current trend is to develop in vitro models able to evaluate PK parameters of drugs considering their physico-chemical properties and by mimicking anatomical and physiological factors. Such models are being increasingly employed in the preclinical stage during drug discovery and development; they can also be used to establish in vitro–in vivo correlations and in quality control procedures. A typical approach is to combine in vitro data with in silico results to build a model whose performance has to be tested in animal experiments [101
As an example, the USP4 dissolution apparatus has often been used to test the dissolution of poorly soluble drugs and sustained-release tablets [102
]; meanwhile, in vitro methods based on freshly explanted animal corneas/eye bulbs mounted on diffusion cells are generally employed to evaluate transcorneal and transcleral drug permeation [104
]. Even though these model systems are very useful for initial screening, they have limitations in that they consider neither the human eye compartmentalization nor the mass transfer process induced in the vitreous humor by the outer aqueous flow. From this perspective, Liu and Wang developed an ex vivo method, which allows for the evaluating the precorneal residence time of ophthalmic formulations. This method consisted in a freshly excised rat cornea placed on a chamber perfused with normal saline solution through two precision pumps controlling the in/out flow rates [107
To simulate the physiological condition, in addition to the perfusion, an efficient model might also mimic the vitreous motion induced by saccadic eye movements. In a recent work, Repetto and colleagues [108
] presented a model consisting in a spherical chamber able to rotate at a scheduled time. Based on this apparatus, Bonfiglio et al. [109
] developed a magnified scale model of the vitreous chamber suitably modified to evaluate the aspects of drug distribution in the vitreous.
A test method simulating the in vivo situation, called the Vitreous Model, was developed by Loch and colleagues to observe the release behavior of ophthalmic dosage forms, such as intravitreal injections or implants, and the consequential drug distribution in the vitreous body [110
]. The same group developed a multi-layer diffusion cell with the purpose of simulating some of the critical parameters that influence drug release and distribution from dosage forms injected in the subconjunctival or intrascleral region [111
Greater insight is provided by the work of Awwad et al. [112
]—these authors proposed an in vitro eye model adapted to human size, consisting in two compartments assembled in such a way as to reproduce the aqueous flow and the mass transfer through the anterior route. This model, named PK-Eye, was employed to assess the PK parameters (clearance, residence time, release profiles, etc.) of therapeutic proteins, such as ranibizumab, bevacizumab and triamcinolone acetonide suspension, and resulted in a promising preclinical tool to study novel long-lasting therapeutics targeted to the posterior segment.
Recently, this model has been refined by our research group [113
], which designed and developed a three-compartment ocular flow cell. Similarly to the PK-Eye model cited above, this new model can be filled with simulated vitreous (agar and hyaluronic acid-based); in addition, it contains a semipermeable disk between the central and the posterior section that can act as a support for retinal cells (Figure 2
In the literature, several papers report the transport behavior of intravitreally administered small and large molecules; results are often obtained by in vivo and ex vivo experiments performed for a few hours. In the case of intravitreally injected protein formulations, it is helpful to follow the stability as a function of time. Accordingly, Patel and colleagues [114
], to predict the stability of three intravitreal monoclonal antibodies, developed ex vivo static, semi-dynamic and dynamic models called ExVit.
With the ExVit static model, it was observed that a significant precipitation and aggregation of proteins, probably due to pH level change, occurred in the vitreous humor (VH) after isolation. The semi-dynamic model consisting in two compartments, the VH- and buffer-compartment, effectively stabilized the pH level and facilitated the migration of degradation products. However, the semi-dynamic model did not completely overcome the limitations related to evaluation of long-term protein stability. Therefore, the same researchers designed a dynamic model comprising three diffusion controlling barriers (two membranes and a gel-matrix) to modulate the diffusion rate of macromolecules. More recently, the same group proposed an ex vivo intravitreal horizontal stability model employed to assess the long-term stability of a bi-specific monoclonal antibody (mAb) named ExVit-HS. It consisted of a two-compartment dynamic model (VH- and buffer-compartment) separated by a diffusion controlling membrane (MWCO of 50 kDa) [115
]. The vitreous-compartment was filled with VH isolated from porcine eyes with an incision placed near the conjunctiva. The results suggested that the ExVit-HS model can be considered a valuable tool for evaluating long term stability of protein drugs and of other therapeutic molecules that are intravitreally injected.
As previously reported, the fate of the drugs following systemic injection or IVI is influenced by the presence of BRB and RPE permeability. However, the information about this topic is currently controversial. Leena Pitkänen et al. [116
] provided permeability values of RPE-choroid as a function of molecule size and lipophilicity. They employed an in vitro diffusion apparatus consisting of a vertical diffusion chamber in which a bovine RPE-choroid was blocked. The permeation data were determined both inward (choroid-to-retina) and outward (retina-to-choroid). Previously, Steuer and colleagues [117
] developed a protocol to isolate the porcine BRB in a rapid and gentle way and explain how to immobilize the intact tissue in a two-chamber polycarbonate device. This RPE model showed a large permeability dynamic range, proving to be a valuable tool for research of BRB drug penetration.
Over recent years, some cell culture models mimicking ocular barriers have been proposed and are considered as a useful alternative to in vivo toxicity tests to investigate pathological conditions and the toxicological screening of compounds [118
]. However, these models are affected by several intrinsic restrictions, mainly because they are formed by cell monolayers grown on a two-dimensional (2D) culture scaffold, which does not take into account the behavior of cells in the three-dimensional (3D) curved native ocular tissue. Therefore, ophthalmic research is strongly interested in developing 3D ocular in vitro models for toxicity testing, safety screening and evaluating long-term drug effects [119
]. 3D models are more suitable to create a cell-based platform whose responses will be as representative as those obtained in in vivo conditions.
Recently, Postnikoff et al. [121
], in the attempt to reproduce the curved cell growth conditions, cultured human papillomavirus-immortalized cells on a curved Millicell-HA membrane. They obtained a stratified, curved epithelial model and employed it to study the biocompatibility of benzalkonium chloride, a preservative widely employed in commercial eye drops. The results underlined the suitability of this model for biocompatibility experiments.
Nowadays, most of the in vitro ocular models described in this review and summarized in Table 3
are in the pipeline and need further investigation.
Computational modeling can also provide information about drug distribution within the vitreous of animals and humans after injection or release by implants [122
]. However, these mathematical models are limited, as they neither consider nor discuss the stability and interactions of proteins/excipients with the vitreous humor components.
Aapo Tervonen et al. [125
] developed a computational model of the physical barrier function of the outer BRB, reflecting the corneal model of Edward and Prausnitz [126
] and aiming to relate the properties of the molecule, such as the lipophilicity and radius, to the permeability of the material and to the tissue diffusion pathways. They also introduced a tight junction model structure for the epithelial model.