Drug-eluting contact lenses: Progress, challenges, and prospects Open Access

The structure of an eye includes anterior (cornea, conjunctiva, iris, ciliary body, lens, and aqueous humor) and posterior (sclera, choroid, retina, and vitreous) segments.¹ Both sections can suffer from various diseases. Glaucoma, cataracts, inflammation, injury, trauma, and dry eye can happen in the anterior segment, while diseases involved in the posterior segment are mainly age-related macular degeneration and diabetic retinopathy.^2–4 If not treated properly, these diseases could even lead to blindness.

Drug-based therapies, i.e., systemic, intraocular, periocular, and topical drugs, represent vital treatments for ophthalmic diseases (Fig. 1).^5,6 Systemic administration is mainly aimed at posterior segment therapy. Its patient compliance is high. However, this delivery is usually hindered by a blood-ocular barrier. Although high-dose administration can reach the desired therapeutic concentration, it would carry the risk of systemic side effects. Periocular and intraocular (intracameral, intravitreal, subretinal, intrastromal, suprachoroidal, and intrastromal) injections enable drugs to bypass some anatomical barriers and drug elimination mechanisms.⁷ Periocular drug delivery (posterior lateral sclera, retrobulbar, peribulbar, and subtenon) was all used in clinical practice for anesthesia during ocular surgery, except for the most popular subconjunctival injections for drug delivery to the eye’s anterior segment (local anesthetics and anti-inflammatory drugs). Periocular delivery also failed to provide sufficient drugs due to the limitation of periocular space, blood retina barrier, and choroidal circulation.⁸ In addition to the pain caused, repeated intraocular injections may lead to complications related to permanent vision loss, e.g., hemorrhages, retinal detachments, cataracts, and infections.⁹ Other options for intraocular drug delivery were inserts and implants, entailing invasive surgical intervention.^10,11

In comparison to oral or intravenous drugs that can pose systemic side effects as well as periocular and intraocular injections that can cause serious complications, topical eye drops are the preferred therapy. They are responsible for >90% of all marketed ophthalmic drugs.¹² Noninvasive topical eye drops were the most popular method of administration for anterior segment disease treatment that typically acts on the cornea, conjunctiva, sclera, and iris. Following absorption by the cornea, the drug could reach tissues of the posterior segment.¹³ The conjunctiva covers nearly 95% of the ocular surface affording sufficient surface area for drug absorption. Drugs in the conjunctiva diffuse through the sclera or cornea or reach the vitreous humor and retina after being cleared into the systemic circulation.¹³ Due to the undesired efflux pumps in the conjunctiva and the dense vascular network, most drugs were applied via the corneal route.¹⁴ 

Notably, topical drug delivery was limited by both the anatomy and the physiology of the eye and dynamic barriers such as reflex lacrimation, continuous blinking, and nasolacrimal duct drainage. These lead to a drug bioavailability of merely 1%–5%.^15–18 Therefore, frequent dosing was necessary to achieve the desired therapeutic concentration. Unfortunately, this resulted in poor patient compliance and increased side effects.^19–22 Cyclodextrins (CDs), gel formulations, colloidal systems, nanoformulations, microneedle patches, and ocular inserts can be applied to increase drug retention time on the ocular surface.²³ However, high-viscosity formulations are irritating to the eyes of many patients. They do not provide accurate dosing and cause blurred vision. Cyclodextrins have limited extremely hydrophobic drug-loading capacity and may not be able to accommodate the release of high doses of drugs. The stability of gel formulations and colloidal systems may be affected by environmental factors such as temperature, pH, and light. The preparation of nanoformulations is more complex and requires controlling parameters such as particle size, shape, and surface properties. The application of microneedle patches appears to require specialized manipulation skills and training. The use of ocular inserts may require specialized physician handling or supervision.

Drug-eluting contact lenses (DECLs) provide a promising approach to managing ocular diseases,²⁴ leading to controlled and continuous drug delivery and enhanced comfort. Contact lenses (CLs) without refractive error were normally applied to correct refractive errors or to protect the eyes. They were composed of three-dimensional polymer networks that were natural drug depots. They have been used for drug delivery recently.²⁵ DECLs significantly prolonged drug residence time (ten times higher than eye drops), improved drug bioavailability (ten times higher than eye drops), as well as diminished dosing frequency, drug concentration, and side effects.^26,27 In addition, DECLs could be removed if treatment required termination. Nevertheless, conventional DECLs showed low drug loading and rapid release within 1–3 h.

To address these concerns, diverse materials were explored to prepare DECLs with sufficient drug-loading capacity and controlled release properties. Different drug-loading and release strategies were also proposed depending on the active component of the drug. Much effort was made in their biosafety evaluation and drug release measurement that mimic the complex physiological environment of an eye, on condition that the efficacy of DECLs is not sacrificed. This paper first reviews the fabrication of DECLs (materials used to prepare CL, drug-loading techniques, and drug release mechanisms) and their in vivo and in vitro characterization (drug release, permeation, and biosafety evaluation). Then, the current challenges facing DECLs are summarized. To conclude, potential future directions for DECLs were proposed to provide inspirational guidelines for their commercialization and clinical applications.

Many materials present the potential for fabricating DECLs due to their excellent physicochemical properties (e.g., biocompatibility, biodegradability, high chemical reactivity, mechanical strength, transparency, and permeability) and biological activity (e.g., promotion of tissue repair and regeneration), including protein-based materials, natural polymeric materials, and synthetic hydrogel materials (Fig. 2).

Bovine serum albumin (BSA) has been widely explored as a drug stabilizer and carrier attributed to its good water solubility, nontoxicity, biodegradability, and chemical stability.³² BSA/Ag porous films combined the antimicrobial properties of Ag nanoparticles and the favorable physicochemical properties of BSA. They can promote cell migration, which makes them well-suited as advanced biomedical materials.³³ BSA also contains several functional groups serving as active sites for subsequent chemical grafting, making BSA/Ag films a perfect candidate for CL. In ophthalmic investigations, hyaluronic acid (HA) was utilized as a wetting agent for DECLs to reduce lysozyme adsorption and denaturation of deposited lysozyme.³⁴ Sodium hyaluronate was proven to not only enhance cell adhesion and proliferation of corneal epithelial cells but also protect cells from oxidative damage as well as downregulate inflammatory cytokines. Wu and co-workers constructed a BSA/Ag/HA composite membrane-based DECL by bonding HA to BSA/Ag films via a simple chemical cross-linking, which prolonged the retention time of HA on the corneal surface [Fig. 2(a)].²⁸ The BSA/Ag/HA film showed diminished corneal opacification and neovascularization and reduced inflammatory response during healing.

Biopolymers were derived from natural sources containing polysaccharides (e.g., chitosan) and peptides (e.g., gelatin and silk fibroin).³⁵ Due to their nontoxicity, biocompatibility, minimal inflammatory response, high oxygen permeability, optical transparency, high wettability, and great chemical stability, they were also considered as candidates for DECL preparation.³⁶ 

Chitosan (CS), a natural polycationic linear polysaccharide derived from the deacetylation of chitin, has excellent ocular tolerance.³⁷ NH₂ of CS could strongly interact with the anionic group of mucins, resulting in an increase in drug residence time.³⁸ Conventionally prepared CS hydrogels suffered from poor light transmission and low mechanical strength³⁹ and, therefore, require additives. Gade et al. prepared DECLs with 100% light transmission by adding plasticizers [glycerol and polyethylene glycol (PEG)] to CS and loaded CLs with moxifloxacin and dexamethasone.³⁸ Quaternary CS, silver nanoparticles, and graphene oxide (GO) were also used to prepare DECLs (loaded with voriconazole), making them antibacterial and antifungal [Fig. 2(b)].²⁹ The lens enables sustained release of the drug and significantly improves fungal keratitis within 7 days.

Gelatin was an extracellular matrix hydrolysis product of collagen. It can be transformed into hydrogels after various types of cross-linking, such as physical, chemical, and enzymatic cross-linking.³¹ It was deemed as an emerging biomaterial for the repair and regeneration of damaged corneal tissue.⁴⁰ However, the poor mechanical strength discouraged its application as DECLs for ophthalmic disease treatment.³¹ It is well known that the mechanical properties of hydrogels could be improved with cross-linking agents. The introduction of polyethylene glycol diacrylate (PEGDA) to gelatine conjugated with methacrylate (GelMA) increased cross-linking density and reduced the biodegradation rate [Fig. 2(d)].³¹ GelMA with a 10% PEGDA concentration achieved minimal swelling (368.6 ± 8.1% vs 520.0% ± 0.0% without PEGDA) and prolonged drug release (50.1 ± 1.3% after 7 days vs 65.6 ± 0.7% release without PEGDA within only 4 days due to degradation). The gelatin hydrogel/CL composites also effectively improved the poor mechanical properties and promoted corneal wound healing. Rutin-coated gelatin hydrogel/CL composites could be obtained by in situ radical polymerization and carboxymethylcellulose/N-hydroxysulfosuccinimide cross-linking reactions [Fig. 2(c)].³⁰ Such DECLs showed an extended release of rutin for up to 14 days and a healing rate of 98.3% ± 0.7% at 48h after surgery.

Regenerated silk fibroin (RSF), produced from the degumming of mulberry cocoons and the dissolution of silk cellulose, consists mainly of glycine, alanine, and serine.^41,42 RSF films possessed good wound-healing properties.⁴³ RSF could be mixed with other components (e.g., CS) and loaded with hydrophilic diclofenac sodium (DS) to fabricate DECLs.⁴⁴ They exhibited visible light transparency of >90%, Young’s modulus of >1.5 MPa, and elongation at the break of >50%, meeting the visual, stiffness, and flexibility requirements of DECLs.⁴⁴ DS loading was elevated with the increasing RSF content of DECLs. They demonstrated a significantly prolonged drug release of up to 11 h at levels above therapeutic concentrations.

Hydroxyethyl methacrylate (p-HEMA) was first used to synthesize soft CLs in the 1960s (Fig. 3). The first case of silicone hydrogel (Si–H) CL was polymerized in 1998 (Fig. 3), making a major breakthrough in the innovation of CL materials. Si–H CL lowered the negative impact on ocular surface tissues and expanded the market of soft CLs.⁴⁶ These two hydrogels, p-HEMA-based hydrogel and Si–H, have been adopted for ophthalmic management.

p-HEMA-based hydrogels, an effective platform for drug delivery, featured great safety, biocompatibility, chemical stability, nontoxicity, and nonantigenicity.⁴⁷ In addition to cross-linkers such as ethylene glycol dimethacrylate (EGDMA) and trimethylene glycol dimethacrylate (TEGDMA), p-HEMA-based CLs are often supplemented with various hydrophilic monomers such as N-vinylpyrrolidone (NVP) and methacrylic acid (MAA) to enrich their water content and enhance their oxygen permeability for comfortable wearing [Fig. 2(e)].⁴⁸ The drug or drug-loaded carriers could be coated onto the p-HEMA-based CLs. A hydrogel prepared from HEMA and MAA monomers, photoinitiator, and cross-linkers (trimethylolpropane trimethacrylate, TMPTMA) was coated with CS that interacted with three types of negatively charged drug nanocarriers (EGCG@DPPC, thiolated lHA@β-1,3-dextran, and thiolated hHA@SB431542) to load drugs.⁴⁹ The designed therapeutic hydrogel enables controlled multistage drug release for sequential treatment of multiple corneal diseases caused by abrasion. However, p-HEMA-based CLs were susceptible to bacterial adhesion due to tear protein deposition, therefore compromising therapeutic efficacy. Hydrophilic chains or groups, e.g., polyethylene glycol (PEG), phosphorylcholine, polyethylene oxide, amphoteric polymers, CD, and HA, were incorporated to improve the antiprotein adsorption of p-HEMA-based CLs.⁵⁰ β-CD-HA was attached to p-HEMA-based hydrogels (p-HEMA/β-CD-HA hydrogels) using polyethyleneimine (PEI) as a cross-linking agent, and the hydrogels were transformed from hydrophobic (WCA = 91°) to hydrophilic (WCA = 52°).⁵¹ The protein deposition of p-HEMA/β-CD-HA decreased by one-third (19.1 μg/cm²) in comparison to that of p-HEMA-based hydrogel (30.3 μg/cm²). Taking advantage of the encapsulation of the drug by β-CD, p-HEMA/β-CD-HA hydrogels obtained a 15% burst release at 1h and a 73% cumulative release at 72h.

Si–H CL was mainly fabricated using silicone materials, e.g., polydimethylsiloxane (PDMS), tris(trimethylsiloxy)silyl propyl vinyl carbamate (TRIS-VC), 3-[tris(trimethylsiloxy)silyl] propyl methacrylate (TRIS), or other silicone macromonomers and hydrophilic monomers, e.g., HEMA, NVP, or N,N-dimethylacrylamide (DMA) [Fig. 2(e)].^52,53 A variety of drugs or compounds have been successfully loaded into Si–H CL to impart interesting properties, for example, vitamin E for UV blocking and hydroxypropylmethylcellulose for drug slow release.⁵⁴ Although hydrophilic soft CLs with a larger swelling ratio demonstrated a higher total drug loading than Si–H CLs, they are not suitable for long-term wear due to low oxygen permeability. Si–H CLs are more popular in clinics since they have higher oxygen permeability.

CLs prepared with different monomers, monomer ratios, and different polymerization methods vary in physicochemical properties and human comfort. As shown in Table I, there are currently three generations of Si–H CLs.^55,56 The most commonly used ones for bandage CLs in clinical diagnosis and treatment are Senofilcon A, Balafilcon A, and Lotrafilcon A.⁵⁷ Clinical trials on DECLs have been conducted (Table II) with emphasis on allergic conjunctivitis, dry eye, corneal epithelial damage, postrefractive surgery antifibrosis, glaucoma, and other areas. The advances attained would serve as references for loading various ophthalmic agents into CL and lay the foundation for clinical applications of DECLs. Recently, one-component hydrogels based on physical cross-linking, e.g., polyvinyl alcohol (PVA),⁵⁸ were also used to prepare DECLs. The reasons for the variation of drug release durations are the differences in the composition of tear fluid and tear simulant, as well as different pathological processes involving different drug binding sites and different CL receptor materials, which need to be optimized in vivo and in vitro experiments before clinical trials. Since current clinical trials involving DECLs usually use the immersion method, the problem of short duration of drug release still exists. Therefore, researchers have proposed biomimetic strategies (environment-responsive stimulation, biosensing) as notable tools for optimizing drug delivery systems, which could meet clinical therapeutic needs by optimization of loading, retention of the drug in the CL structure during storage, modulation of drug release once applied to the eye, and maintenance of CL physical properties.⁵⁹ 

A significant amount of research has been devoted to loading drugs into CL. Numerous approaches, including traditional soaking, molecular imprinting, micro-, and nanoparticle carriers, multilayers, and embedded implantation, have been developed.

Johnson & Johnson® has developed a drug-loading contact lens called ACUVUE® Theravision™ that incorporates the antihistamine ketotifen into a synthetic contact lens. Two phase 3 clinical studies have shown positive treatment results in patients with allergic conjunctivitis, with a significant reduction in eye itching within 15 min and for up to 12 h of wearing the contact lens.

Soaking is the simplest way to prepare DECLs. The drug molecules diffuse from the solution into CL during the swelling of the immersed CL.⁶⁰ 0.45% ketorolac eye drops could be used for postoperative cataract pain management.⁶¹ CLs were soaked in their solution and used after transepithelial photorefractive keratectomy (TransPRK) to relieve ocular surface pain.⁶² Soaking the CL with 0.1% diclofenac sodium drops was also proven effective in reducing pain and discomfort after TransPRK.⁶³ However, conventional soaking methods typically lead to low drug loading as well as undesired swelling and optical properties of the CL due to poor drug solubility.⁶⁴ Moreover, DECLs prepared by this method could not perform continuous drug delivery beyond a few hours,⁶⁵ and its burst release dramatically lowered drug bioavailability. In addition, not all drugs could diffuse into the CL via simple immersion.

Several techniques were developed to maximize the drug loading of DECLs after immersion and to extend the release time. Electrostatic interactions between the comonomers of CLs and drugs are considered an advantage in addressing this problem. For example, the addition of negatively charged comonomers (acrylic acid, MAA, and 4-methyl-4-pentenoic acid) to p-HEMA-based CLs caused a significant increase in the loading of ofloxacin (18 times) and neomycin (53 times) at pH 6.5 [Fig. 4(a)].⁶⁶ Positively charged ionic monomers, such as methacrylamidopropyl trimethyl ammonium chloride and 2-methacryloyloxyethyl phosphate, have also been shown to improve the ability of CLs to load and release oppositely charged drugs using ion-exchange reactions. In these cases, however, the CL may show changes in swelling rates due to the adsorption and release of charged drugs,⁵² which, in turn, affect the optical properties. Alternative methods included the addition of ionic surfactant substances and ion-exchange resins containing sulphonic acid groups, carboxyl groups, quaternary ammonium groups, etc. during the preparation of CLs.^69–72 Among them, surfactant introduction could resist protein absorption due to increased hydrophilicity.⁵² 

For hydrophobic drugs that are difficult to load, incorporating compounds that interact hydrophobically with the drug can be a solution. Graphene oxide not only improved the water-holding capacity and optical properties of the CL but also markedly increased cyclosporine uptake (7.4 ± 1.0 μg, p < 0.01). This is due to the more swelling of the CL than the CL without graphene oxide and the hydrophobic interaction (π-σ stacking) between graphene oxide and cyclosporine [Fig. 4(b)].⁶⁷ The pH and temperature of the drug-loading solution also play an important role in drug loading.⁷³ The hydrogel tended to swell more quickly at higher pH, providing more drug storage space.⁷⁴ Mild heating (60 °C) could facilitate the loading of antibiotics (moxifloxacin) and anti-inflammatory drugs (diclofenac and ketorolac) into CLs polymerized mainly by HEMA and methyl methacrylate. High temperature would lead to the relaxation of the polymer chains and the heat-absorbing interaction of the drug with the polymer.⁷⁵ Dipping Si–H CLs in organic solvents (e.g., n-propanol) caused a high swelling organosilicon network, accelerating the loading process of hydrophobic drugs remarkably.

Nanoaggregates of vitamin E could be added into CLs as a diffusion barrier for hydrophobic and hydrophilic drugs [Fig. 4(c)].⁶⁸ Vitamin E also inhibited keratinocyte apoptosis after photorefractive keratectomy (PRK) surgery and prevented cornea damage by UV radiation. Typically, CLs were soaked in ethanol containing vitamin E.⁷⁶ Their swelling enabled vitamin E to distribute and bind to the long-chain polymer units of the substrate. The pirfenidone release from unmodified CLs could last approximately 20 min to reach a 90% release, while loading with 20% and 40% vitamin E increased that duration to approximately 80 and 260 min, respectively.⁷⁷ The amount of drug reaching the cornea was approximately 50 times that of the corresponding eye drops. In addition, the dexamethasone (DX) release time of commercial CLs (ACUVUE OASYS™, NIGHT&DAY™, and O®2OPTIX™) would be extended to 7–9 days after being loaded with 30% vitamin E, representing a 9–16-fold increase from pristine CLs.⁷⁸ It should be noted that DECLs containing vitamin E frequently displayed first-order “burst” kinetics, and this needed to be optimized to avoid toxicity while maintaining clinically relevant therapeutic doses. Additionally, lipophilic vitamin E nanoaggregates may alter the mechanical properties and ionic permeability of CLs, possibly making them impractical to wear.⁷⁹ 

Molecular imprinting is a technique that involves manipulating the hydrogel structure to create a higher affinity for the target drug. In detail, the template molecule, i.e., the released drug, was polymerized with functional monomers and cross-linkers that could interact with it by ionic, van der Waals, and hydrogen bonding interactions [Fig. 5(a)].⁸² Monomers and cross-linkers were selected to mimic the interactions between the drug and the target receptor in vivo. Once polymerized, the unreacted monomer and template molecules are extracted, leaving the high affinity pockets to achieve drug loading [Fig. 5(a)]. This technique has been used to improve drug uptake and drug release kinetics.^83,84 The drug-loading capacity of soft CLs prepared via molecular imprinting was two to three times greater than that of CLs made by conventional methods.⁸⁵ Acrylic acid and benzyl methacrylate (BzMA) were selected to add into p-HEMA-based CL to form a binding cavity that mimics the opioid growth factor receptor binding site for naltrexone (NTX) [Fig. 5(b)].⁸⁰ AA’s weak interaction with the aliphatic nitrogen, hydroxyl, and carbonyl groups of NTX as well as BzMA’s correct orientation in the imprinted cavity enhanced the affinity of the drug for the binding region. Compared with nonimprinted CLs, treated CLs allowed for more NTX loading and were capable of controlling NTX release. The release was continuous for at least 2 days in a well-stirred bulk medium and for longer periods in dynamic conditions that mimic physiological conditions, maintaining therapeutic concentrations in the lacrimal fluid until day 3.

Interestingly, molecular imprinting could also endow the structured, colored hydrogels with active sites responsive to target drug molecules.⁷⁴ The binding or unbinding behavior of the target drug molecule induces changes in the volume and refractive index of the hydrogel matrix that can be translated into readable optical signal shifts by the photonic crystal structure. Deng et al. proposed the concept of a molecular imprinted structural color CL, combining the molecular imprint (timolol interacting with the carboxyl site of MAA) and the inverse opal structure [Fig. 5(c)].⁷⁴ Such DECLs could effectively convert drug release behavior into a readable signal through their color change, which remained unaffected after five reuses.

In addition, molecular imprinting was combined with other strategies to afford control release of two or more drugs from DECLs. For example, AA was first introduced into Si–H CL to improve MXF release through ionic interaction or hydrogen bonding to the drug molecule [Fig. 5(d)].⁸¹ Then, layer-by-layer (LbL) deposition of polyelectrolytes (alginate, polylysine, and HA) was performed on the CL surface and cross-linked with 1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide hydrochloride, setting a barrier for the diffusion of diclofenac sodium (DCF) [Fig. 5(d)].⁸¹ Such DECLs released DCF and MXF with concentrations above the IC50 (half maximal inhibitory concentration) and minimum inhibitory concentration values of S. aureus and S. epidermidis for 9 and 10 days, respectively.

The composition of the CL determined its affinity for a given drug, so each imprinted DECL may require a specific development approach. When multiple drugs are required for treatment, the design and scaling up of molecular imprinted DECLs could be challenging. The engagement of various components may cause a negative impact on the optical and physical properties of the lenses. In addition, the growing number of functional monomers available for the preparation of imprinted DECLs implied that more in vivo studies on their safety would be necessary.

Micro-nanoparticles were being extensively investigated as promising carriers for loading drugs into DECLs. Surface interactions of micro-nanoparticle surfaces with drugs were appealing. Nontoxic gold nanoparticles (GNPs) with high surface area tend to adsorb drugs on their surfaces, substantially increasing the amount of drug in DECLs.⁸⁶ Importantly, their penetration in all layers of the cornea would be gradually removed. DECLs were obtained using two routes: (1) GNPs absorbed the drug before being exposed to CL and (2) GNPs absorbed the drug from the solution after mixing with the CL.⁸⁷ The latter was found to yield the largest load of timolol (346 ± 10 μg).

Drugs could also be encapsulated into micro-nanoparticles, which were then incorporated into the CL or bound to the CL surface. In these cases, the drug first diffused out of the dispersed particles and then through the lens to the tear film. The continuous release of the drug from the DECLs extended over days or weeks. Particles used for drug encapsulation included microemulsions, nanoparticles, polymeric micelles, and liposomes. A p-HEMA-based DECL was designed by dispersing nanoparticles (PEG as a hydrophilic outer shell, p-HEMA as a hydrophobic inner shell, and PCL as a hydrophobic core) encapsulating anti-inflammatory drugs into the CL [Fig. 6(a)].⁴⁷ The drug release of DECLs containing 7.5% nanoparticles amounted to about 80 μg/d and could last 12 days, which was within the therapeutic range. Propoxylated glycerol triglyceride (PGT) nanoparticles have also been applied to the load drug (timolol) followed by binding to the prefabricated CL.⁹⁰ Because of the slow hydrolysis of the ester bond linking timolol to the PGT, DECLs released timolol in phosphate buffered saline (PBS) for approximately one month at room temperature. Unfortunately, the low transparency of CLs caused by some nanoparticles limited their application.

It was found that this downside would disappear when the NP size is less than 180 nm and exacerbate above 300 nm.⁹¹Smaller micelles would be preferable for loading into CLs. Latanoprost- and timolol-loaded mPEG-PLA micelles prepared by thin-film hydration were embedded in CLs [Fig. 6(b)].⁸⁸ The ultra-small particle size and narrow particle size distribution of the produced micelles provided desirable transmission of DECLs. The constructed DECLs were able to release timolol and latanoprost in simulated tears for 144 and 120h, respectively, and in rabbit tears for up to 120 and 96h, respectively. Compared with eye drops, timolol and latanoprost release based on DECLs demonstrated notable improvements in residence time (79.6 and 122.2 times) and bioavailability (2.2 and 7.3 times).

Liposomes, i.e., bilayer lipid vesicles composed of phospholipids and cholesterol, were known to enhance corneal permeation and reduce drug toxicity.^89,92,93 Liposomes were one of the most common drug delivery carrier particles in the field of ophthalmology. Hydrophilic drugs could be encapsulated in an aqueous center, and hydrophobic drugs could be positioned in the region of the hydrophobic tail [Fig. 6(c)].^94,95 The commonly used phospholipids encompassed phosphatidylcholine, phosphatidylethanolamine, dipalmitoylphosphatidylcholine, distearoylphosphatidylcholine, dipalmitoylphosphatidylglycerol, dioleoylphosphatidylethanolamine, etc.⁸⁹ Through different liposome preparation methods, liposomes with variable sizes and structures can be obtained, such as small unilamellar vesicles (SUVs), large unilamellar vesicles (LUVs), giant unilamellar vesicles (GUVs), and multilamellar vesicles (MLVs) [Fig. 6(c)]. The tight binding of cholesterol to lipids controlled the elasticity and membrane fluidity of the lipid bilayer, and the lipids increased the permeability toward biological membranes.⁹⁶ As opposed to the 2–5 min residence time after ciprofloxacin drops, DECLs loaded with drug-encapsulated liposomes (prepared with soya lecithin and cholesterol) reached up to 24 h of residence time.⁹⁷ The sustained release time of multilayer liposomes containing ciprofloxacin was longer than that of monolayer liposomes.⁹⁷ In addition, liposomes could be modified with surface charges, polymer chains, antibodies, and proteins to improve their stability both in vitro and in vivo.⁹⁸ Another merit of liposomal particles was that the large water center and the exterior of the lipid bilayer allowed the incorporation of large molecules (drugs, peptides, proteins, plasmid DNA, antisense oligonucleotides, or ribonucleases).^98,99

Another emerging method of prolonging CL drug release is the fabrication of multilayer lenses.¹⁹ Additional polymeric layers or implants encapsulating the drug were sandwiched between the hydrogel layers, preventing burst release of the drug.¹⁰⁰ Dexamethasone-polylactic acid-hydroxyacetic acid copolymer (PLGA) films were embedded in a soft CL (Methafilcon) in order to construct a delivery system (Dex-DS) [Fig. 7(a)].²⁵ It showed sustained drug release for one week. The retinal drug concentration of rabbits wearing Dex-DS was 200 times higher (per hour) than that of Dex drops, and they showed lower systemic (serum) dexamethasone concentrations. Notably, the above technique permitted multiple drug delivery. Three implants were polymerized by the mixture of monomers (Irgacure 184, EGDMA, DMA, NVP, siloxane, and HEMA) containing timolol, bimatoprost, and HA, which were then placed in a mold for further polymerization with the monomer mixture to obtain DECLs [Fig. 7(b)].¹⁰¹ The transmission of the CL was not affected since implants were placed at its periphery. In vitro drug release data from these DECLs revealed a longer sustained release (72 h) and lower burst release compared to soaking and direct drug polymerization CLs. In a rabbit model, it increased the duration of IOP reduction to 120h, while a peak-to-valley distribution was observed after treatment with eye drops.

Microneedle technology was a minimally invasive means of topical drug delivery offering local, long-lasting, and potent ocular treatment with good patient compliance.¹⁰⁶ There are three main types of microneedles, i.e., hollow, solid coated, and polymer soluble microneedles.⁹¹ Microneedles were feasible as drug reservoirs to control drug release for ocular disease treatment, especially for keratitis and corneal neovascularization (CNV). CLs equipped with cyclosporine A microneedles (height and base diameter 535 ± 15.7 and 287 ± 2.00 μm, respectively) were prepared by the micromolding technique [Fig. 7(c)].¹⁰² Due to the water solubility of the main component (polyvinylpyrrolidone) of the microneedle, they completely dissolved in the cornea after 60 s of application. The flux (9.04 ± 0.88 μg/cm²/h) and retention (81 ± 5.1 μg/g) of cyclosporine A in the excised porcine cornea were higher than commercially available Cyclomune® eye drops (flux of 2.11 ± 0.56 μg/cm²/h and retention of 53 ± 1.1 μg/g) after 5 min. Cyclosporine A was detected in the cornea (45.5%), atrium (29.1%), lens (8.3%), vitreous body (5.2%), sclera (3.0%), and choroid-retinal ring (0.2%). Microneedle and CL combination has been proposed for biphasic drug release or the use of multiple compartments in microneedles to pack multiple drugs for synergistic treatment.¹⁰³ An eye-contact patch containing a double layer of MNs combined the advantages of HA and cross-linked methacrylated HA (MeHA) [Fig. 7(d)]. The highly soluble HA acted as an inner core of MNs, providing stiffness and fast release. MeHA was used as an outer shell layer that is more resistant to dissolution, achieving slow drug releases.¹⁰³ However, the specific sensitivity of the cornea (corneal pain) and corneal damage triggered by larger-sized microneedles (at the submillimeter size level) limited the clinical implementation of microneedles in human eyes. Silicon nanoneedles (Si NNs) caught researchers’ attention due to their nanoscale size, easy surface functionalization, low toxicity, and slow biodegradation.^107,108 Biphasic drug release was achieved by embedding biodegradable Si NNs on the surface of CL prepared from tear-soluble polyvinyl alcohol [Fig. 7(e)].¹⁰⁴ Initially, a rapid release of anti-inflammatory drugs (less than 1 min) from the CL is followed by a long-term sustained release (more than one month) of therapeutic CNV drugs from Si NNs (hydrolysis of Si NNs to silicic acid and hydrogen gas by reaction with tears). Surface porosity or size of Si NNs could be adjusted to control the drug-loading capacity. The minimal amount of bevacizumab needed to use Si NNs in the rabbit corneal model (about 1.5 μg) was much lower than eye drops or subconjunctival injections (usually >1.2 mg). These DECLs are shown to be the most effective in treating CNV on days 1–7 and continue to work until day 28.

Except for micro-nanoneedles, additional micro-nanopores became new drug reservoirs for CLs. A team applied a spherical mold manufacturing process and soft lithography to fabricate a CL with embedded μ-tubes (both CL and μ-tubule layer were made of PDMS) [Fig. 7(f)].¹⁰⁵ It provided a dynamically responsive antiglaucoma therapy in which drug release is activated by stretching the CL due to changes in the patient’s IOP. It also featured simple, noninvasive, and extended drug release up to 45 days. In addition, DECLs could adjust the dose in real time to avoid possible toxicity and adverse manifestations.

Once DECLs were soaked in the drug solution, the drug molecules diffused into the network of CLs to achieve the same drug concentration between the solution and the hydrogel phase.¹⁰⁹ If a drug failed to interact or bind to the CL polymer network, the drug would diffuse extremely fast when DECLs are applied on the ocular surface.¹⁰⁹ The overall release mechanism of DECLs was generally based on molecular diffusion, though innovative techniques have been developed to improve drug release from CLs.¹¹⁰ Molecular diffusion drove drug release from the CL to the tear film behind the lens and subsequent absorption by the cornea. The diffusion of the drug to the ocular surface was influenced by the molecular weight and water solubility of the drug, the pore size, and the thickness of CLs, as well as the proportion and the size of other materials incorporated into CLs.

Drug release could be aided by external stimuli such as light, pH, temperature, ionic strength, or enzymes of the surrounding environment. Light-mediated drug release systems are more ideal for ocular applications to disease therapy, in that the eye is the sole organ permitting light penetration.¹¹¹ Mechanisms of light-mediated drug release involved photo cracking of chemical bonds between polymers and drugs (e.g., o-nitrobenzyl, coumarin, and pyrene), photoisomerization reactions (e.g., azobenzene and spiro-pyran), and photothermal reactions [e.g., GNPs and poly (N-isopropyl acrylamide)].¹¹² The NHS-ester of caged timolol was reacted with 2-aminoethyl methacrylate, and the product was subsequently copolymerized with HEMA, EGDMA, and NVP to obtain a DECL.¹¹³ It underwent an excited state photoisomerization reaction during indoor daylight exposure, leading to the cracking of carbonate bonds and the release of timolol. Daylight CL passively released timolol at a dosage up to 4305 times the inhibitory concentration of βARs between successive blinks over a 10-h period. More importantly, the concentration of timolol released into the eye is ∼10 000 times higher than that of a single eye drop, and the total timolol amount is only ∼5.7% of a single eye drop. A red-shifted cage crosslinker would make it possible to release drugs from lenses under indoor lighting systems (e.g., fluorescent lamps and LEDs).

The difference in pH among the packaging solution (pH < 7), saline used for cleaning (pH = 7.4), and patient’s tears (pH > 7.8 in dry eyes)¹¹⁴ might aid the design of drug release systems. DECLs containing cyclosporine-loaded Eudragit S100 nanoparticles [Fig. 8(a)] were developed.¹¹⁴ Eudragit S100 displayed dissolution behavior at pH above 7, while it prevented drug release during sterilization of the packed solution and a monomer extraction step at pH below 7.¹¹⁹ The initial burst release of the DECLs with integrated S100 nanoparticles (3.7 ± 0.85 μg) was found to be significantly lower than the lens directly loaded with the drug (12.01 ± 0.38 μg), indicating a controlled release from Eudragit S100 nanoparticles.¹¹⁴ In addition, the drug was continuously released for up to 156 h. A blended membrane of ethyl cellulose and Eudragit S100 was perforated and used as a DECL inner layer loaded with diclofenac sodium [Fig. 8(a)]. Its drug release was negligible in a PBS at pH 6.8 and sustained in the tear solution for more than 24 h.¹¹⁵ Weakened ocular immunity and CL abrasion could lead to corneal infections, which further changed the microenvironment considerably, such as a decrease in pH (from 7.4 to 5.5).¹¹⁶ The antibiotic (gentamicin sulfate) was reversibly grafted onto the PEI pretreated CL via a Schiff base reaction [Fig. 8(a)].¹¹⁶ The chemical bond formed would keep the drug in a neutral environment (pH 7.4, PBS) and release it in an acidic environment. The dynamic chemical bond between the polyphenol hydroxyl group and the phenylboronic acid has also been shown to trigger drug delivery by the acidified environment described above.¹²⁰ 

When the temperature acts as a switch for drug release, it should usually be set between 25 and 34 °C, corresponding to the ambient and ocular surface temperatures, respectively. Timolol bases were loaded in nanoparticles prepared from PGT and EGDMA and subsequently polymerized with a mixture of HEMA monomers to form p-HEMA-based DECLs.¹¹⁷ The -OH group in the timolol molecule reacted with the polymer network through a nucleophilic reaction to form an ester, followed by hydrolysis for drug release [Fig. 8(b)]. The duration of drug release shortened and the total amount of drug released increased with the rising temperature [Fig. 8(b)].

The rich electrolytes present in the tear fluid might serve as a basis for the development of an ion-sensitive drug delivery system. The DECLs were prepared using a cellulose acetate film, containing betaxolol hydrochloride (BH)—ion-exchange resin (polystyrene-divinylbenzene sulfonate resin) composite as the inner layer and a silicone hydrogel as the outer layer [Fig. 8(c)].¹¹⁸ There was an ionic interaction between the sulfate group of the resin and the NH group of BH. In simulated tear fluid, 93.6% BH was released within 6h and DECLs continued to be released for 12h, whereas only <3% of BH was released in distilled water. Additionally, DECLs were stable in distilled water at 5 °C for at least 30 days without significant change in drug release kinetics. In contrast, the drug was released in rabbit tears for more than 168 h and the corneal residence time significantly improved.

The human tear film also contained higher concentrations of proteins including lysozyme, albumin, immunoglobulin, and β-lactoglobulin.⁵¹ Certain drug NP carriers have been reported to release drugs by lysozyme degradation. Two CLs were obtained by incorporating CS-pAA NPs in PVA hydrogels, as well as by adding in situ gelled NPs and cellulose nanocrystals (CNCs) in PVA lenses.¹²¹ These polymer NPs were degraded by hydrolytic cleavage of CS chains in 0.2 mM lysozyme. In terms of drug release, NP-PVA showed an extended release of 28 h, and NP-CNC-PVA lenses exhibited greater extended-release potential. PEI coated nanodiamonds (NDs) were cross-linked with an enzymatically scavengable CS to form an ND-nanogel bearing timolol maleate, which was embedded in p-HEMA-based CL.¹²² Both drug-soaked and molecular imprinted CLs demonstrated almost complete release of timolol maleate in 1 h, while the cumulative drug release from ND-nanogel lenses after lysozyme treatment was 9.41 μg within 24 h.

In order to examine whether DECLs could achieve the therapeutic concentration, in vitro drug release assessment was usually conducted first. The sinking environment of the eye should be simulated in in vitro release experiments, since concentration gradients exerted a measurable effect on the release process.^123, In vitro release assays were classified into three main types: (i) release under static conditions in a container (beaker, bottle, tube, etc.); (ii) release under fluid flow in a designed microfluidic chamber; and (iii) release under blink-simulated conditions.¹⁰⁹ The conventional static determination method involved dipping DECLs in a bottle containing the release medium and stirring at 34 °C (ocular surface temperature). Certain samples were removed and immediately replaced with an equal volume of release medium at predetermined time intervals to prevent drug saturation. Among de-ionized water, PBS solution, 0.9% NaCl solution, and artificial tear (ALF), ALF was the most suitable drug release medium due to its similarity to tear. Actually, the solubility of the drug, the volume of the release medium, and the mixing rate are all factors that should be considered.¹²⁴ The poorer drug solubility caused by small volumes would result in an overestimation of the CL’s ability to regulate drug release. If large volumes of release media were selected or frequently changed to fresh release media, the sinking effect along with wide concentration gradients would probably accelerate the release process.¹²⁴ 

Microfluidic chambers were able to adjust media flow and total volume, which could better simulate the release environment under the physiological conditions of the eyes.⁴⁸ Currently, three major types of microfluidic devices are used for in vitro release studies [Figs. 9(a)–9(c)].⁴⁸ The first microfluidic device consisted of a microfluidic chip, a syringe pump, a slide, an inner chamber, and two fluidic channels. The drug release from DECLs was extended by several weeks in this microfluidic device compared to the same DECLs in a beaker.¹²⁷ However, the device was unable to simulate the cornea and eyelids, or alter tear film thickness by evaporation or blinking. The 3D printed eye model, a useful tool for measuring the release,¹²⁸ was applied to another microfluidic device. It is designed to enable the flow of fluid to simulate a physiological tear flow state. The third device had the smallest internal chamber volume (45 μl, close to the maximum tear film volume in the eye) and multiple outlets converged in a single collector.¹²⁹ Although this device was more similar to natural ocular physiology, it still did not take into account the blink mechanism. Therefore, the eye model was covered with a bovine cornea and a blinking platform was added by using a gear system and an electronic system [Fig. 9(d)].¹²⁵ The Simublink device was also used to simulate the blinking motion during DECL wearing. A 3.5 g PMMA cylinder was used to slide over the hydrogel at a speed of 14 cm/s with a 2 s pause between “blinks” and a 16 kPa pressure applied to mimic the eyelid effect.¹³⁰ The drug release from p-HEMA-based DECLs increased with loading pressure and friction cycling. However, the influence was negligible when the drug was encapsulated in the liposomes of the CL.¹³¹ 3D printing techniques were also employed to construct an entire ocular model device to assess the effects of air exposure, flow rate, and blink frequency on drug release.^126,132 This model contained a lower eyelid component that provided structural support for the eye. The component also helped keeping the DECLs in a stable position [Fig. 9(e)]. Because of its thin structure (500 μm), the upper eyelid was able to blink over the lower eyelid. The lower eyelid was also designed to accumulate fluid and let the fluid drip into a special collection tray.

Some ideal release profiles were difficult to reproduce in vivo due to the significant differences in in vitro and in vivo release conditions. Tissue or animal models were considered to be more reliable and representative. Thus, it is necessary to evaluate the in vivo performance of the drug delivery system. Rabbit eyes are similar in size and structure to human eyes and were widely used as a model.¹³³ In detail, sterile DECLs were carefully placed on rabbit cornea without anesthesia (left eye as the control, right eye wearing a DECL). Tear fluid was collected from the conjunctival sac at set time intervals using disposable glass capillaries. The collected fluid was then treated with methanol to precipitate the proteins. After shaking and centrifugation, the supernatant was collected and analyzed for drug content.

Efforts devoted to DECLs have paid off in the treatment of anterior segment diseases. However, anatomically complex barriers in the eye still prevent effective drug delivery. Overcoming these barriers to achieve the desired drug concentration at the therapeutic site is a key objective in the design of noninvasive drug delivery systems. Herein, we describe the potential barriers to drug diffusion and possible strategies to overcome them.

Four key ocular barriers challenge effective drug delivery in the eye, namely, tear film barrier, corneal barrier, blood-aqueous barriers, and blood-retinal barriers (Fig. 10).¹³⁴ The tear film, consisting of a lipid layer, an aqueous layer, and an underlying mucosal layer, nourished and protected the cornea and has a volume of approximately 7–10 μl. The drug was lost in advance partly due to tear overflow, while the rest would be diluted in the tear film and would drain rapidly through the nasolacrimal duct. The lipid layer resists evaporation of water as well as reduces drug absorption into the cornea and sclera.¹³⁵ In the aqueous layer, drug interactions with enzymes (e.g., lysozyme), mucins, and proteins (e.g., albumin) could lead to nonspecific binding, which precludes further drug penetration. The mucin layer on the epithelial surface was a shield for pathogens. Its charge and viscoelasticity might hinder drug diffusion.¹³⁶ In turn, some drugs were subjected to the adhesion of the mucus layer, showing a longer retention time and improving drug absorption in ocular tissues.

The avascular cornea was a more efficient route for ocular drug penetration. It is composed of five layers: epithelium, Bowman’s membrane, stroma, Descemet’s membrane, and endothelium. The basal cells of the hydrophobic epithelial layer were continuously renewed and flattened when pushed upward to form tight junctions. Such tight junctions are the main barrier for drug transcorneal permeation after topical administration.¹³⁷ Three methods were exploited to permeate the epithelium: (i) the transcellular pathway; (ii) the paracellular pathway (the pore size between epithelial cells is about 2.0 nm, allowing molecules <500 Da to pass); and (iii) the transporter protein-mediated pathway. Bowman’s membrane (composed mainly of type I and V collagen) with relatively large pores (approximately 10 μm) was not considered a strong barrier.¹³⁸ The stromal layer accounts for ∼90% of the corneal thickness and is composed primarily of water and collagen (mainly type I). The hydrophilic-hydrophobic alternation between different corneal layers also constituted a great hindrance to the passive diffusion of drugs.¹³⁹ Dua’s layer lies anterior to Descemet’s membrane in the cornea and is 10 μm thick (mainly type I and some type VI collagen).¹⁴⁰ Descemet’s membrane (mainly type IV collagen) with 38 nm mean pore size would prevent macromolecules and particles that directly entered the stroma from reaching the endothelium. The endothelium is a nonrenewing monolayer of cells, and this monolayer provides a small barrier to the penetration of hydrophilic particles. Although it also owns tight junctions, it is less confined to drugs than the epithelium due to greater permeability and discontinuity.

In addition, the diffusion of drugs across the cornea into surrounding intraocular tissues and vitreous humor is impeded by aqueous humor clearance and degradation due to intracellular metabolic enzymes.¹⁴¹ The opposite direction of aqueous humor movement and drug molecule diffusion is another dynamic barrier.¹⁴² After passing through the aqueous phase, drug binding to the melanin in the ciliary body and iris also reduces the fraction of drug reaching the posterior segment.⁸ Because of the negatively charged HA and glycosaminoglycan proteins in the vitreous, the charge and the lipophilicity of the drug or its carrier strongly influence its diffusion behavior.¹⁴³ Following the vitreous, the inner limiting membrane with an average pore size of 10–25 nm is a structural barrier to molecular diffusion into the retina.¹⁴⁴ If the drug primarily targeted the retina and vitreous, both the choroid and Bruch’s membrane would prevent its diffusion.¹⁴² The dense blood supplied in the choroidal layer and the narrow pores of the choroidal endothelium result in the significant removal of small molecule drugs to the systemic circulation.

The blood-aqueous barrier and the blood-retinal barrier arose mainly after systemic drugs. The former is constructed by endothelial cells of the iris vessels and tight junctions of nonpigmented ciliary epithelial cells of the ciliary body, inhibiting the movement of molecules from the iris vessels. The latter is formed primarily by two main structures: external tight junctions of retinal pigment epithelial cells and intercellular junction’s retinal capillary endothelial cells. Under pathological (neovascular disease) conditions, higher concentrations of the vascular endothelial growth factor disrupt the tight junctions of the retinal pigment epithelium, promoting greater extracellular diffusion of hydrophilic and larger molecules.¹⁴⁵ A variety of enzymes and efflux proteins produced by retinal pigment epithelial cells formed a metabolic barrier to drugs.¹⁴¹ The latter two barriers are not the focus of this paper.

Here, we summarized the possible strategies for drug delivery of DECLs through the ocular barrier based on biochemical or physical methods.

Many organic compounds could assist drugs in overcoming certain corneal barrier layers by altering corneal permeability.¹⁴⁶ This provided an insight into the design of DECLs. Some organic compounds opened the tight junctions of the corneal epithelium to penetrate the stroma. The superficial epithelium of the cornea is surrounded by Ca²⁺-dependent cell-binding sites.¹⁴⁷ The effect of the epithelial barrier is determined by the concentration of Ca²⁺ ions available to the plasma membrane.^147,148 Polyamino carboxylic acids [ethylenediaminetetraacetic acid (EDTA), ethylenediamine-N,N′-disuccinic acid (EDDS), and ethylene glycol-bis(2-aminoethylether)-N,N,N′,N′-tetraacetic acid (EGTA)] exhibited a superior ability to form complexes with Ca²⁺, demonstrating better drug (riboflavin) permeability (penetration into the corneal stroma) than the PBS control [Fig. 11(a)].¹⁴⁷ In addition, benzalkonium chloride, Brij 78, and borneol also contribute to penetration enhancement, with benzalkonium chloride being the most effective one.^153,154 However, high concentrations of penetration enhancers might cause irritation to the corneal tissue.

The use of a carrier-based transcytosis pathway and paracellular pathway allowed for effective corneal tissue penetration. Polyamino acid-based S-nitrosothiols with high cationic charge density acted as NO carriers to bypass the hindrance of corneal tissue, and the drug was diffused via their super-cation induced transcytosis [Fig. 11(b)].¹⁴⁹ In the wild-type mice model, the carrier was observed in the entire corneal and the trabecular reticulum in the anterior chamber. This system integrated super-cation and responsive glutathione to enhance intraocular delivery of topically administered drugs, effectively relieving high IOP of mice with glaucoma. Currently, NPs with sizes <200 nm were recognized to penetrate the corneal epithelium due to their high surface area and ocular residence time.¹⁵⁵ The size and the surface charge of the NPs were further optimized in order to be transported into corneal deeper layers where disease severely affected vision (e.g., neovascularization of the stroma). Carbon quantum dots (CQDs, ζ = + 32.5 mV) were nanoscale. They exhibited good water solubility, tunable surface functionality, and low cytotoxicity.¹⁵⁶ CQDs of size <3 nm and charge >20 mV were not internalized by the cells and could help macromolecules (dextran at 150 kDa) to penetrate into the corneal endothelium [Fig. 11(c)].¹⁵⁰ 

NPs were also used to improve drug delivery by enhancing adhesion to ocular surface mucins through adsorption, diffusion, and charge mechanisms [Fig. 11(d)]. For adsorption, hydrogen bonding and van der Waals forces dominated the interaction between mucin and NPs. NPs with a surface chemical structure similar to that of PVA, HEMA, and poly (methacrylic acid) could strongly hydrogen bond to mucin.¹⁵⁷ The diffusion of polymer chains on NPs into mucous membranes is also responsible for adhesion. Polymers characterized in long, linear, and flexible polymeric chains (e.g., PEG) were previously shown to improve interpenetration with mucin.^151,158 Nanocarriers, including solid lipid nanoparticles, nanolipidic carriers, cubosomes, niosomes, liposomes, nanomicelles, and spanlastics, were identified for drug delivery targeting diseases of the posterior segment.¹⁵⁹ Amphiphilic surfactants endowed these carriers the stronger ability to diffuse through the lipophilic and hydrophilic corneal barriers. When loaded with dexamethasone, they permeated via trans- and paracellular pathways, excluding spanolastics and niosomes via paracellular pathways. Among them, nanolipidic and nanomicelles carriers exhibit better corneal permeability.¹⁵⁹ Polysaccharide-based NPs, such as CS, CS/HA, and alginate NPs, bound to the mucin through electrostatic interactions and prolonged drug release.¹⁶⁰ Moreover, CS was reported to open tight junctions of the corneal epithelium.^161,162 In addition to these physical mechanisms, covalent bonds or specific affinity interactions were applied to promote adhesion. Thiol groups could form disulfide bonds with cysteine-rich mucus glycoproteins and were often introduced on carriers.^163,164 Thiolated β-CD not only improved the drug solubility but also increased the permeability of sodium fluorescein in the conjunctiva, sclera, and cornea by 9.6-fold, 7.1-fold, and 5.3-fold, respectively [Fig. 11(e)].¹⁵² The phenylboronic acid-modified nanocarriers were found to form reversible covalent complexes with 1,2-diols or 1,3-diols on mucin.¹⁶⁵ 

Furthermore, four methods have been developed to facilitate the penetration of the mucin layer [Fig. 11(f)]. In addition to the aforementioned introduction of a layer of highly hydrophilic and near-neutral surface-charged polymer (PEG) on the drug carrier surface, zwitterionic surfaces (net neutral surface charge due to equally balanced positive and negative charges) minimized the affinity to mucin.^166,167 The third method was to impart surface potential switching properties to the drug carrier. The initial weak negative charge allowed them to penetrate the mucus. Then, they were converted to a positive charge to facilitate epithelial endocytosis.^168,169 Such charge switching was typically achieved by enzymatic cleavage of charged substrates.¹⁷⁰ In the last case, drug carriers equipped with enzymes such as papain or thiomers cleaved disulfide bonds in the mucin network^164,171 and facilitated penetration into the epithelium. To maintain the protective and lubricating function of the mucus layer, only the mucins adjacent to the particles were selectively cleaved. The latter two promising approaches have not been widely reported in the eyes due to the absence of enzymes capable of cleaving the negative charges of nanocarriers and the instability of free thiols.

Biomolecules mostly penetrate ocular tissues via transcellular pathways and transport protein-mediated pathways. The enhanced adhesion and endocytosis of corneal cells had the potential to override the epithelial physiological barrier.¹⁷² Cell-penetrating peptides, typically short-chain peptides with a positive charge, were capable of penetrating cell membranes. Peptide-based molecular hydrogels featured biocompatibility, biodegradability, and minimal immunity. Cationic self-assembled peptide-based hydrogels that interact with negatively charged mucins showed better cellular uptake, leading to enhanced corneal penetration [Fig. 12(a)].¹⁷³ Cationic hydrogels could also interact with transmembrane mucins (i.e., MUC1, MUC4, and MUC5) via ionic interactions, to furnish long-term retention on the corneal surface.

Novel peptides with protein transduction properties are attractive for challenging ocular tissue barriers. The peptide GGG(ARKKAAKA)₄, MW = 3.5 kD, has a protein transduction domain. It functioned as a peptide for ocular delivery (POD) and could deliver both small molecules (including fluorescent probes and siRNA) and large molecules (including plasmid DNA and quantum dots) to cells and murine eye tissue in vivo [Fig. 12(b)].¹⁷⁴ The permeation of topically administered POD throughout the sclera and the choroid was likely to allow the delivery of drugs or genes to choroidal vessels and endothelial cells, whose proliferation is related to various ocular posterior segment diseases. Topical administration of POD-bound drugs to the cornea would also deliver drugs to the corneal epithelium and the anterior chamber.

Electric fields can enhance drug delivery. In electroporation, high-voltage pulses were applied to create transient pores in lipid bilayers to introduce molecules into the cells. Recent advances in microfabrication allowed the introduction of electrodes and cells into CLs. Ideally, electroporation devices should be equipped with nontoxic and flexible electrodes that fit the physiological shape of the human body. For example, metal-polymer conductors (MPCs) were prepared using PDMS and liquid metal (eutectic gallium-indium, EGaIn) and then patterned into the CL (MPCC) [Fig. 13(a)].¹⁷⁵ Compared to weak fluorescence in rabbit stroma treated with free siRNA (gene silencing of VEGF-A), more fluorescence (representing siRNA for treating CNV) was detected in the corneal stroma of the MPCC-siRNA-treated group. Moreover, vessel length, circumference width, and area in the former group were much lower than those in the latter group (0.47 ± 0.02 cm vs 0.71 ± 0.05 cm; 5.91 ± 0.83 vs 8.48 ± 0.43; 1.75 ± 0.24 mm² vs 3.80 ± 0.45 mm²).

Iontophoresis promotes tissue permeability utilizing low voltage and facilitates ionic or nonionic drug crossing the barriers through electromigration and electroosmosis.^177,178 For ophthalmic treatment, iontophoresis could drive the ionic drug delivery to the aqueous humor and the vitreous humor. When a voltage is applied, current flows from the anode into the tissue and then back to the cathode. Christopher and Chauhan placed two circular electrodes on commercial CLs containing Nile blue and fluorescein as hydrophobic and hydrophilic drug analogs, respectively [Fig. 13(b)]. The electrodes were connected to a power supply set to the desired current.¹⁷⁶ The absorption of Nile Blue increased after the current application (0.125–0.250 mA), and its amount was a function of the action time of the electric field and the magnitude of the current applied. Similar results were observed for fluorescein. Importantly, the permeability of the dye increased in the presence of an electric field. Despite these promising results, further research should be carried out to refine electrode materials and electrode positions to minimize the impact on the view and to reduce manufacturing costs.

An attractive aspect of micro-nanoneedles is their ability to deliver drugs precisely to the depth of the cornea that needs to be treated. Most of them came with the capacity to create microchannels that provided a pathway through physiological barriers without apparent injury or pain.¹⁷⁹ Solid microneedles, coated microneedles, hollow microneedles, and dissolving microneedles have been proven to overcome barriers and offer drugs for the treatment of glaucoma, age-related macular degeneration, and infectious corneal disease.¹⁸⁰ Among them, the dissolving microneedles, made from water-soluble polymers, could be rapidly dissolved by water in the tear film and other ocular tissues, significantly reducing the sustained time and improving clinical acceptance. The microneedle patches prepared were able to bypass the tear film washout and the corneal epithelial barrier to reach the corneal stroma. As reported, the human corneal thickness is 527 ± 28 μm and the microneedle lengths used in the test models were 500 μm (mouse), 552 μm (rabbit), and 480 μm (pig).¹⁸¹ 

Diffusion cells for drug permeability measurements were usually based on the transwell system or the transwell-snapwell system.¹⁵¹ The transwell diffusion system included a membrane-separated donor and a receptor as well as a central compartment. Briefly, the cornea excised from an animal’s eye was installed in a compartment between a DECL-added donor and an ATF-added recipient. Similar to the above, a certain volume was removed from the recipient compartment and replaced with an equal amount of ATF at a selected time point. The diffusion of the drug or carrier toward the receptor zones was quantified to determine its permeability. The transwell-snapwell system was composed of a donor and recipient room, split by a central chamber containing a vertical mucus layer, which was surrounded by two permeable filters connected to a snapwell ring that allowed particles to permeate through the mucus.¹⁸² 

The main advantage of fluorescence imaging is that it could be manipulated in in vivo and in vitro models. Confocal laser scanning microscopy (CLSM) qualitatively assessed penetration by applying a fluorescent filter to excite a specific fluorophore. The fluorescence intensity emitted was directly related to the amount of drug/carrier labeled by a fluorescent dye. CLSM also offered a z-directional resolution of the fluorescent signal that can be used to more visually assess the drug penetration into the tissue.

Current research on DECLs covers blinding ophthalmic diseases such as keratoconus, glaucoma, retinal vascular diseases, etc. However, few studies were conducted to confirm their effectiveness and safety both in vivo and in vitro. DECLs were not commercially available on a large scale. The ketotifen CL developed by Johnson & Johnson® was the first DECL in the world. It was approved in Japan and Canada in 2021.¹⁰⁹ The biocompatibility of DECLs is a common concern in the field of biomaterials and clinical medicine.

In a review of 24 publications, most in vivo and in vitro studies tended to focus on anterior segment diseases, mainly bacterial infections and noninfection inflammatory conditions (Table III). The duration of drug release varies between in vivo and in vitro experiments, with the duration of the drug release typically reaching about several weeks in vitro, compared to 4–7 days in in vivo experiments. The variation may be due to differences in the composition of the tear fluid. Researchers have already improved the in vitro-in vivo correlations of drug release duration by 3D printing eye models, but physiological parameters such as the tear flow rate and temperature still need to be further refined.¹²⁸ As mentioned above, DECLs could prolong the retention time of drugs on the ocular surface, but it is notable that long-term wear of DECLs could result in corneal tissue hypoxia, the issue of material oxygen permeability should be taken into account in the design of DECLs. In addition to the traditional in vitro cellular assays to validate cytotoxicity, the chicken embryo chorioallantois membrane vascular assay and drug permeability validation in isolated eyes were added. In vivo biocompatibility validation included observation of ocular tissue irritation. Cytotoxicity assays and ocular tissue structure staining have been used to monitor the toxicity of drugs and carriers with a lack of long-term safety observations.¹⁰⁹ In addition, the combined results of these studies provide new ideas for the development of DECLs, including copolymerization, drug loading, and biocompatibility confirmation to optimize the clinical efficacy of ophthalmic agents.

Although DECLs have made satisfactory progress in drug loading and controlled release, certain challenges must be noted and addressed prior to commercialization. (1) Key properties, such as water content, mechanical properties, ion permeability, light transmission, and oxygen permeability, tended to change unfavorably during drug loading using the strategies described above.²⁷ For example, insufficient oxygen permeability would cause corneal hypoxia, leading to dry eyes, blurred vision, and even ulcerative keratitis.⁷² (2) The complicated manufacturing process of DECLs resulted in higher costs.⁶⁶ (3) Tear proteins deposited in CLs lead to microbial attachment and inflammation. (4) Emerging nanotechnology-based strategies for improving ocular drug delivery suffered from drug leaching during storage, drug loss during lens sterilization, and instability of nanoparticles. (5) Environmental stimulus-responsive drug release strategies hold promises for on-demand drug delivery. However, for the ocular surface, excessive variations in pH, temperature, and ionic strength might make it less realistic for ophthalmic applications. In addition, minor differences in the amount of enzyme activity in tears made it difficult to develop nanomaterials that are sensitive to such modest changes. (6) Multilayer and embedded implant DECLs presented the advantage for applications to a wide range of diseases. However, the PLGA frequently used could degrade to acidic monomers, increasing toxicity.²⁰⁰ (7) The poor deformability and the rigidity of the electrodes typically used in DECLs based on electric field technology preclude them from fitting in the shape of the target area, leading to inefficient transportation and erratic efficacy. (8) In addition to the excessive complexity of the manufacturing process, after application, the mechanical strength of microneedles (especially polymeric microneedles) was reduced by tissue moisture, possibly leading to several complications. (9) DECLs were mostly still limited to the animal level. There is a huge gap between animal studies and clinical applications.

With deepening insights into the fields of material science and biology, the methodology library for developing DECLs is becoming more abundant. In order to realize efficient drug packaging, precise dose delivery, sustained release, sufficient bioavailability, and better patient compliance, DECLs were always being investigated. DECLs would also reduce the risks associated with wear (keratitis, corneal erosion, dry eye, conjunctivitis). Nevertheless, the vast majority of DECLs required more research in order to gain commercial acceptance. For example, increasing the bioavailability of DECLs without compromising the safety of formulation is imperative.

Currently, smart drug release is a hot topic. Some attempts have been made to integrate stimulus-responsive nanoparticles with CLs to minimize drug loss during storage and to accelerate release in response to light, ocular temperature, pH, or enzyme triggers. More efforts are necessary for improving the drug delivery system due to the strict limitations and instability of the triggering conditions in the eyes.

The combination of DECLs with biosensing technology is a new trend that not only allows noninvasive and continuous monitoring of various basic physical/biochemical parameters²⁰¹ but also delivers drugs for the treatment of ocular diseases. Up to now, smart CLs have been developed for diagnosing diseases, such as glaucoma^202,203 and diabetes.^204–206 More endeavors will appear in integrating sensing units on CLs with drug delivery systems. Aided by artificial intelligence, these smart CLs may be able to select the appropriate treatment and therapeutic concentration of drugs based on the data obtained from monitoring. Given the multifunctional integration of contact lenses, their safety and shelf life deserve special attention.

This work was supported by the National Natural Science Foundation of China (NNSFC) (Nos. 82171032 and 21773022), the Science Research Project of The Third People’s Hospital of Dalian (Nos. 2022ky001 and 2022ky002), the Liaoning Provincial Basic Applied Research Program (No. 2022JH2/101300036), and the Dalian Science and Technology Innovation Fund project (No. 2023JJ12034).

The authors have no conflicts to disclose.

Ethics approval is not required.

  Dongdong Gao and Chunxiao Yan contributed equally to this work.

Dongdong Gao: Conceptualization (equal); Data curation (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal). Chunxiao Yan: Conceptualization (equal); Data curation (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal). Yong Wang: Data curation (equal); Investigation (equal). Heqing Yang: Data curation (equal); Investigation (equal). Mengxin Liu: Data curation (equal); Investigation (equal). Yi Wang: Data curation (equal); Investigation (equal). Chunmei Li: Writing – review & editing (equal). Chao Li: Writing – review & editing (equal). Gang Cheng: Conceptualization (equal); Methodology (equal); Writing – review & editing (equal). Lijun Zhang: Conceptualization (equal); Funding acquisition (equal); Methodology (equal); Project administration (equal); Resources (equal); Supervision (equal); Writing – review & editing (equal).

The data that support the findings of this study are available within the article.