Ultra‐thin, high strength, antibiotic‐eluting sutures for prevention of ophthalmic infection

Abstract Sutures are applied almost universally at the site of trauma or surgery, making them an ideal platform to modulate the local, postoperative biological response, and improve surgical outcomes. To date, the only globally marketed drug‐eluting sutures are coated with triclosan for antibacterial application in general surgery. Loading drug directly into the suture rather than coating the surface offers the potential to provide drug delivery functionality to microsurgical sutures and achieve sustained drug delivery without increasing suture thickness. However, conventional methods for drug incorporation directly into the suture adversely affect breaking strength. Thus, there are no market offerings for drug‐eluting sutures, drug‐coated, or otherwise, in ophthalmology, where very thin sutures are required. Sutures themselves help facilitate bacterial infection, and antibiotic eye drops are commonly prescribed to prevent infection after ocular surgeries. An antibiotic‐eluting suture may prevent bacterial colonization of sutures and preclude patient compliance issues with eye drops. We report twisting of hundreds of individual drug‐loaded, electrospun nanofibers into a single, ultra‐thin, multifilament suture capable of meeting both size and strength requirements for microsurgical ocular procedures. Nanofiber‐based polycaprolactone sutures demonstrated no loss in strength with loading of 8% levofloxacin, unlike monofilament sutures which lost more than 50% strength. Moreover, nanofiber‐based sutures retained strength with loading of a broad range of drugs, provided antibiotic delivery for 30 days in rat eyes, and prevented ocular infection in a rat model of bacterial keratitis.

loses strength as the surrounding tissue heals and gains strength, and is able to scale to commercial production. We sought to develop sutures that meet each of these requirements while also providing sufficient and controlled release of a therapeutic moiety in order to improve surgical outcomes.
Sutures are a promising means of therapeutic delivery directly to the surgical site. However, clinical implementation of such technology has been limited due to the inability of drug-loaded sutures to meet United States Pharmacopeia (U.S.P.) standards for suture strength. 6,7 Conventional suture manufacturing processes, such as melt extrusion, are not compatible with many therapeutic moieties, and drug-loaded sutures in preclinical development have demonstrated breaking strength of only 10% of clinical specifications. 4,[7][8][9][10][11][12][13][14][15][16] Drug-coated sutures have been developed to circumvent these shortcomings, but can be limited in their ability to meet diameter requirements, load sufficient drug, control drug release, and/or scale manufacturing. 17 There is a significant need for an antimicrobial suture in ocular surgery, where, globally, more than 12 million procedures per year use conventional nylon sutures to close ocular wounds and incisions. 24 Nonabsorbable nylon sutures are a mainstay of ocular surgery due to their biocompatibility and strength retention at the surgical site. 25 Nylon sutures are used in procedures such as penetrating keratoplasty, where sutures remain in the eye for 12 to 24 months, and as with other implantable devices, increase the risk of infection following ophthalmic procedures due to their susceptibility to bacterial adhesion, proliferation, and biofilm formation. [26][27][28][29][30][31] Incidence of infectious keratitis following penetrating keratoplasty has been reported between 1.76% and 12.1%. 32 Suture-related complications are implicated in 20% to over 50% of these cases, and can have devastating consequences, including poor visual outcomes, reintervention, and graft failure. 28,[32][33][34] Thus, it is particularly important to provide for local antibacterial functionality along with implantation of nonabsorbable sutures within the eye. Local antibiotic delivery from the suture itself would provide bacterial inhibition at the vulnerable surgical incision and help alleviate concerns of noncompliance with topical antibiotic eye drops, which are often prescribed postoperatively.
Antibiotic-eluting sutures may also reduce the need for postoperative oral antibiotic prescriptions, the systemic administration of which can lead to emergence of resistant organisms and associated complications such as Clostridium difficile infection and life-threatening diarrhea. 35,36 In addition to keratoplasty, glaucoma, retinal detachment, vitrectomy, and cataract surgeries, where nylon sutures have been used, antibiotic-eluting sutures may also decrease the risk of infection associated with concurrent implantation of keratoprostheses, lacrimal stents, orbital plates, glaucoma drainage implants, or other ocular devices. [37][38][39][40][41][42][43][44] An antibacterial suture for ophthalmology must be extremely fine (20-50 μm in diameter; U.S.P. sizes 10-0 through 8-0) while retaining high strength for the duration of the intended application and providing sufficient release of an antibiotic agent to reduce or prevent ophthalmic infection. 6,7 Here, we provide the first demonstration of electrospinning of drug-loaded nanofibers that are then twisted in a controlled manner to form ultra-thin, high strength, drug-eluting sutures of appropriate breaking strength and diameter to meet U.S.P. specifications for ocular surgery. 8,9,13,14 In order to provide an antibacterial alternative to the use of nylon sutures in ocular surgery, we manufactured sutures composed of polycaprolactone (PCL) and levofloxacin (Levo). Levo is a third-generation fluoroquinolone and broad-spectrum ophthalmic antibiotic indicated for treatment of bacterial conjunctivitis. 45 PCL is a biocompatible polymer capable of longterm degradation that has been used in sutures and other medical devices for more than 30 years. 25,[46][47][48] We evaluated antibioticeluting suture size, breaking strength, pharmacokinetics, biocompatibility, and efficacy in a rat model of bacterial keratitis.

| RESULTS
We hypothesized that sutures composed of twisted PCL/Levo nanofibers would provide suitable strength at the surgical site for an extended duration while also delivering antibiotic in a sufficient and controlled manner to prevent postoperative suture colonization and ocular infection.

| Nanofiber suture manufacture and characterization
In order to limit the effect of drug loading on the strength of polymeric matrices, we engineered a novel manufacturing system capable of producing and twisting together hundreds of individual drugloaded, polymeric nanofibers ( Figure 1(a)). High voltage was applied to a polymer or polymer/drug solution pumped at a controlled flow rate in order to form polymeric fibers. However, rather than collecting fibers on a rotating drum, which is often employed in electrospinning applications, fibers were collected in parallel between two grounded collectors situated perpendicularly to the syringe pump. Rotation of one collector results in the twisting of deposited parallel fibers into a single 17-cm-long multifilament suture. The amount of fiber deposition, and consequently, suture diameter was reproducibly tuned by adjusting spray time.
Seventeen kilovolts were applied to a 10% PCL solution in hexafluoroisopropanol (HFIP) flowing at 450 μL/h for 60 s, followed by twisting of deposited fibers 1575 times in order to manufacture a single, multifilament suture. Scanning electron microscopy (SEM) of multifilament sutures confirmed manufacture of a highly uniform, nonporous, and defect-free thread composed of nanofibers ( Figure 1 (b)). Notably, individual nanofibers had a flat, ribbon-shaped morphology, and had an average width of 729.9 ± 246 nm. Multifilament, drug-loaded sutures were cylindrical in nature and met U.S.P. specifications for 10-0 suture diameter (20-29 μm), making them a suitable size for ocular surgery. They were also comparable in both size and shape to commercially available 10-0 Ethilon ® (nylon) sutures ( Figure 1(b)).
The principal challenge for translation of drug-loaded sutures to the clinic has been an inability to meet U.S.P. specifications for suture strength. Thus, we next examined the impact of fiber conformation, drug concentration and type, and suture diameter on suture breaking strength. Figure 2 Next, we explored the amount of drug that could be loaded into multifilament nanofiber sutures while maintaining U.S.P. strength specifications. We produced 1575 twist, 28 μm multifilament nanofiber sutures composed of PCL with no drug (0%) or with 8%, 16%, 24%, or 40% Levo within the suture formulation. PCL sutures with 16% or more Levo had a significantly lower breaking strength (p < 0.05) than PCL sutures alone or with 8% Levo (Figure 2(b)). It was possible to include up to 24% Levo within the multifilament nanofiber suture formulation while still surpassing clinical strength requirements for a 10-0 suture. Notably, even with inclusion of 40% Levo into the suture formulation, multifilament PCL nanofiber suture breaking strength was significantly higher (p < 0.05) than a monofilament nanofiber suture with 8% Levo (Figure 2

| In vivo suture biocompatibility
In order to further evaluate the potential use of multifilament nanofiber sutures as an alternative to nylon or other commercially available sutures, we assessed the local tissue reaction to implantation of 3 × 2 mm long 10-0 nylon, Vicryl ® , PCL, PCL/8% Levo, or PCL/16% Levo sutures after 2 days in the rat corneal stroma. There were no gross signs of irritation, inflammation, or infection among any of the treated or control groups (not shown). Histological analysis ( Figure S1) revealed that implantation of PCL or PCL/Levo sutures did not cause neovascularization, and that the tissue reaction was comparable to commercially available nylon sutures. Notably, a small ring of immune cells was observed surrounding implanted absorbable Vicryl ® sutures ( Figure S1).  To our knowledge, this is the first report of drug-loaded sutures that surpass U.S.P. breaking strength specifications. 4,8,9,13 Similar to prior reports in the literature, micron-sized electrospun PCL monofilament sutures lost more than 50% of their strength upon inclusion of Levo. In contrast, twisted, multifilament nanofiber sutures did not lose strength with inclusion of an equivalent amount of Levo. PCL is a semicrystalline, hydrophobic, and biodegradable polymer. 47 It has been shown that the process of electrospinning alone can enhance nanofiber molecular orientation and that PCL nanofibers increase in tensile strength with reduced diameter due to molecular confinement. 62,63 This phenomenon is not observed in PCL fibers produced via the melt flow extrusion process used to manufacture commercially available sutures today. 62 Further, PCL crystallinity increases along with a decrease in molecular weight. 47 We sought to maximize fiber crystallinity, and consequently suture strength, by electrospinning nanofibers composed of low molecular weight PCL. Moreover, twisting of individual nanofibers into a multifilament suture provided additional structural reinforcement, resistance to breakage, and knot security. 6 The flat, ribbon-shaped morphology of the individual nanofibers suggests that the twisting process led to stretching of nanofibers, which has also been shown to improve polymer chain alignment and tensile strength. 64  structures. This might also lead to the burst release of a fraction of the encapsulated Levo observed following implantation of drug-loaded sutures into rat eyes. We hypothesize that prior to PCL degradation, the drug delivery profile of small molecules from multifilament PCL sutures will depend primarily on the solubility and distribution of the drug in the nanofibers. As such, we anticipate that more hydrophobic drugs will have a slower overall release rate than Levo, but may also provide for burst release in the immediate postoperative period. implanted. [26][27][28][29]32,33,61,[65][66][67] Given the small diameter of 10-0-sized sutures utilized in ophthalmic procedures and the difficulty of manufacturing and coating such sutures, it will be challenging for sutures manufactured via conventional methods to provide sufficient drug delivery over this duration. 19,23 However, local antibiotic delivery from the drug-loaded sutures reported here may preclude issues of poor patient compliance with topical eye drops, prevent suturerelated infections that lead to treatment failure and reintervention, reduce the need for oral antibiotic use, decrease the risk of infection associated with implantable ocular devices, and serve as an alternative to the more than 12 million nylon sutures used in ocular procedures each year. 24,28,32,36,61,68 In order to translate this technology platform for patient use in ophthalmology and beyond, we plan to conduct long-term studies to evaluate suture strength retention and degradation in vivo while also assessing late-stage wound sealing and tissue healing.

| Bacterial inoculation and evaluation
Sprague Dawley rats were anesthetized as described above. The operative eye was then scratched using a 20 G needle (Fisher Scientific) prior to implantation of three 2 mm long nylon (n = 12), Vicryl ® (n = 4), or PCL/8% Levo (n = 4) suture filaments. Nylon sutures are commonly used for corneal transplant and ocular trauma surgeries, while Vicryl sutures are commonly used for cataract procedures. A 100 μL droplet containing 1 × 10 8 CFU/mL of S. aureus was then applied to the ocular surface, left in place for 10 mins, and then removed with a sterile wick without touching the eye. Ten microliter of 0.5% Levo (w/v, the concentration in commercially available eye drops) solution was administered topically either once postoperatively, as would be done by the surgeon, or three times daily, as would be prescribed for prophylaxis, to rat eyes with nylon sutures (n = 4, each). Two days after implantation, gross images were taken of each eye, prior to swabbing the cornea with a cotton-tipped applicator (Fisher Scientific), and streaking onto tryptic soy agar (Fisher Scientific) plates. Plates were stored in an incubator at 37 C for 24 h and then imaged. After swabbing the eye, eyes were enucleated and either prepared for histological evaluation as described above (n = 3 for each condition) or evaluated for bacterial load (n = 4 for each condition).
Briefly, each eye was placed in sterile tryptic soy broth (Fisher Scientific) and homogenized using a Power Gen 125 homogenizer (Fisher Scientific) for 4 min. Samples were then centrifuged at 300 rcf for 5 min, and optical density of the supernatant was measured at a wavelength of 600 nm using a Synergy Mx microplate reader (Biotek, Winooski, Vermont). The bacterial load was determined by subtracting the optical density of fresh tryptic soy broth from experimental values prior to applying a conversion of 0.1 OD to 10 8 CFU/mL. Infection was confirmed by a positive swab culture and bacterial load significantly higher than a nonoperated control eye.
Alternatively, nylon, PCL/8% Levo, and PCL/16% Levo sutures (n = 8, each) were implanted into rat corneas on day 0, with inoculation of S. aureus to only the PCL/Levo suture conditions, as described above. On day 2, rat corneas were swabbed to evaluate infection. On day 5, the corneas of all rats were scratched and inoculated with S. aureus. On day 7, swabs were taken of each cornea followed by either histological evaluation, bacterial homogenization, or removal of sutures for examination via SEM (n = 4 for each condition). For the latter experiment, sutures were removed from the cornea and fixed in formalin (Sigma-Aldrich) for 30 min prior to washing with PBS and dehydration with increasing concentrations of ethanol (Fisher Scientific). Sutures were then imaged by SEM as described above.

| Statistical analysis
Suture breaking strength, Levo concentration, and bacterial load are presented as mean ± standard error. Statistical significance for breaking strength and bacterial load data was determined via one-way ANOVA followed by Tukey test. Statistical significance for the Kaplan-Meier curve of long-term infection prevention was determined via the Mantel-Cox test. Statistical significance is shown as p < 0.05 ( # or *), p < 0.01 ( ## or **), or ***p < 0.001. Outliers in pharmacokinetics data were determined via Grubbs' test at p < 0.01, resulting in removal of three outliers in the D7 data, one each in the aqueous and cornea data in the PCL/Levo 8% group, and one in the cornea data in the PCL/Levo 16% group.